The influx of streaming services has made a dent in broadcast and cable – but far from displaced them, new data shows. CC-licensed photo by Alan Levine on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

It’s Friday, so there’s another post due at the Social Warming Substack at about 0845 UK time. It’s about search and chatbots.

The Overspill is going on holiday for two weeks. Returning June 17.

A selection of 10 links for you. Goggle eyed. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

US slows plans to retire coal-fired plants as power demand from AI surges • FT

Amanda Chu:

»

The staggering electricity demand needed to power next-generation technology is forcing the US to rely on yesterday’s fuel source: coal.

Retirement dates for the country’s ageing fleet of coal-fired power plants are being pushed back as concerns over grid reliability and expectations of soaring electricity demand force operators to keep capacity online.

The shift in phasing out these facilities underscores a growing dilemma facing the Biden administration as the US race to lead in artificial intelligence and manufacturing drives an unprecedented growth in power demand that clashes with its decarbonisation targets. The International Energy Agency estimates the AI application ChatGPT uses nearly 10 times as much electricity as Google Search.

An estimated 54 gigawatts of US coal powered generation assets, about 4% of the country’s total electricity capacity, is expected to be retired by the end of the decade, a 40% downward revision from last year, according to S&P Global Commodity Insights, citing reliability concerns.

“You can’t replace the fossil plants fast enough to meet the demand,” said Joe Craft, chief executive of Alliance Resource Partners, one of the largest US coal producers. “In order to be a first mover on AI, we’re going to need to embrace maintaining what we have.”

Operators slowing down retirements include Alliant Energy, which last week delayed plans to convert its Wisconsin coal-fired plant to gas from 2025 to 2028.

«

That delta between ChatGPT and Google Search is bananas. If Google’s AI Overview is using anything like it.. wow.
unique link to this extract

The Terraformer Mark One • Terraform Industries Blog

»

The Terraformer is a carbon-neutral drop-in successor to drilling for fossil fuels.

The Terraformer is designed to integrate directly with a standard 1 MW solar array. No grid connection, no interconnection queue. The Terraformer gets solar energy to market as energy dense, clean, cheap, carbon neutral synthetic natural gas.

The Terraformer produces 1000 cubic feet of natural gas per hour of operation. It is optimized for 25% utilization, typical for utility scale solar arrays, and in this configuration produces 6,000 cubic feet/day.

Operating the equivalent of 2190 hours per year, one Terraformer produces over 2 million cubic feet of natural gas. At $10/Mcf [million cubic feet] sale price and $54/Mcf for IRA PTCs (45V, 45Q, 45E) each unit produces up to $150,000 of annual revenue.

A gigawatt-scale solar array integrated with 1,000 Terraformers will produce enough natural gas to supply 20,000 homes.

A self-funding global fleet of 400 million Terraformers, rolled out over the next two decades, will provide all of humanity with permanent unconditional energy abundance for the first time in history, completing the mission of the industrial revolution.

Production starts Q2 2024.

«

Ben Thompson interviewed Casey Handmer, the chief executive of this company, at Stratechery. I thought it sounded barmy: why pull carbon dioxide out of the air in order to burn it? But the carbon-neutral approach is interesting. It demands 2,000 litres of water per day. That’s not free, but it’s not that expensive either – about £5 per day. Perhaps it could work. The problem is always scale. One isn’t enough, and a thousand would require a lot of solar power and space – all to produce a version of something that’s otherwise cheap (absent swingeing carbon taxes).
unique link to this extract

US poised to invest millions in mRNA bird flu vaccine amid H5N1 scare • Daily Telegraph via Yahoo

Maeve Cullinan and Paul Nuki:

»

The US government is poised to announce a multimillion-dollar investment in mRNA vaccines for H5N1 bird flu. The move comes as the virus continues to spread in mammals, threatening a new pandemic if it makes the jump to humans.

The US Biomedical Advanced Research and Development Authority (BARDA) is understood to be nearing an agreement with Moderna to finance human trials for its experimental mRNA bird flu vaccine. The deal would include a commitment to stockpile millions of vaccines if the trials were successful, the Financial Times reported on Thursday. 

Senior officials at the World Health Organization (WHO) in Geneva welcomed the news, amid fears the H5N1 outbreak, which has killed millions of animals in the last two years, could yet spread to humans.

Jeremy Farrar, chief scientist at the WHO, said investing in mRNA jabs for H5N1 was “an important step forward”, while Maria Van Kerkhove, head of the emerging diseases and zoonosis unit, said it demonstrated “active pandemic preparedness” on the part of the Americans. The recent progress of H5N1 into US cattle was “concerning,” she added.

H5N1, which has a human case fatality rate of 50%, has been circulating for more than 20 years, causing thousands of human infections and more than 400 deaths primarily in people working with animals over that period.

«

Also: most recent H5N1 case detected in US farm worker shows the virus has mutated in a manner “associated with viral adaptation to mammalian hosts”.

Just a watching brief.

unique link to this extract

OpenAI is rebooting its robotics team • Forbes

Kenrick Cai:

»

Over the past year, OpenAI’s in-house startup fund has invested in several well-capitalized companies trying to develop humanoid robots, including Figure AI (raised $745m), 1X Technologies ($125m) and Physical Intelligence ($70m). It hinted at a possible robotics reboot in a February press release for Figure’s latest fundraise; And one month later, Figure debuted a video of its robot demonstrating rudimentary speech and reasoning skills supported by a large multimodal model trained by OpenAI. “We’ve always planned to come back to robotics and we see a path with Figure to explore what humanoid robots can achieve when powered by highly capable multimodal models,” vice president Peter Welinder, previously a member of OpenAI’s robotics team, said.

Two sources told Forbes that OpenAI intends to coexist rather than compete against such companies, building technology that the robot makers will integrate into their own systems. And the listing notes that engineers hired for the position would be tasked with collaborating with “external partners” as well as training AI models. Sources said it’s unclear whether OpenAI plans to develop robotics hardware, which it struggled to do several years ago. Its widening ambitions have recently been marked by some turbulence — a series of high-level safety team departures and an accusation from actress Scarlett Johansson that it appropriated her voice for its ChatGPT product “Sky.”

The narrower focus for the robotics team this time around may nevertheless have some overlap with companies it hopes to engage in business. Companies like Covariant, started by former OpenAI robotics team members, are also attempting to train their own robotics models. And, two sources said, OpenAI has already gone head to head with companies in this space for a limited pool of talent.

«

Can’t have a robot apocalypse without robots *looks at camera, taps temple*. But at least they’ll strangle you with a flirty laugh.
unique link to this extract

Twelve Angry Trump Jurors • The Washington Post

Alexandra Petri:

»

[Twelve jurors are assembled in a deliberation room while a fan spins lazily overhead. Jurors open windows, fan themselves and make small talk.]

Juror 3: Hot in here.

Juror 11: Hottest day of the year, they say.

Juror 5: Well, we won’t be in here long. Which is good, because I have to go to a lecture about the presidency of Grover Cleveland.

Juror 7: Sounds skippable.

Juror 5: It’s his second term, not his first term.

Juror 1: Okay, gentlemen. We can do this a number of ways. We can discuss and then vote. Or we can take a preliminary vote, see where we stand and then discuss. Our result has to be 12 to nothing, either way.
Juror 3: Let’s do a preliminary vote. Maybe we can all go home.

Juror 1: All right. Show of hands: Who thinks he is guilty?

[Eleven hands go up. All heads swivel to Juror 8, sitting at the end of the table with his hand firmly on its surface.]

Juror 1: No?

Juror 8: I’m just not convinced.

Juror 1: Oh, boy.

«

Petri writes wonderfully funny columns. Enjoy!
unique link to this extract

OpenAI says its tools were used in foreign influence campaigns • Axios

Ina Fried:

»

OpenAI said Thursday that it has seen several foreign influence campaigns tap the power of its AI models to help generate and translate content, but has yet to see novel attacks enabled through its tools.

Supercharging misinformation efforts has been seen as a key risk associated with generative AI, though it has been an open question just how the tools would be used and by whom.

OpenAI said in a new report that it has seen its tools used by several existing foreign influence operations, including efforts based in Russia, China, Iran and Israel.

For example, the Chinese network known as “Spamouflage” used OpenAI’s tools to debug code, research media and generate posts in Chinese, English, Japanese and Korean.

The Russian “Doppelganger” effort, meanwhile, tapped OpenAI models to generate social media content in several languages as well as to translate articles, generate headlines and convert news articles into Facebook posts.

Meanwhile, an Iranian operation known as the International Union of Virtual Media used OpenAI tools to both generate and translate long-form articles, headlines and website tags, while an Israeli commercial company called STOIC ran multiple covert influence campaigns around the world, using OpenAI models to generate articles and comments that were then posted to Instagram, Facebook, X, and other websites.

OpenAI said it also detected and disrupted a previously unknown Russian campaign dubbed “Bad Grammar” that operated on Telegram. The effort targeted Ukraine, Moldova, the Baltic States and the United States and used OpenAI models both to debug code for a Telegram bot and to create short, political comments in Russian and English.

«

unique link to this extract

IRS plans to make its free tax filing program permanent • CNN Politics

Katie Lobosco:

»

The Internal Revenue Service said Thursday that it will continue and expand its free tax filing program in 2025.

A pilot version of the online program, known as Direct File, launched earlier this year. It provides step-by-step guidance to taxpayers filing their federal tax returns.

Direct File was available to people with certain simple tax returns in 12 states during the 2024 tax filing season. More than 140,000 people successfully filed their federal tax returns using Direct File, exceeding the agency’s expectations.

Next year, the IRS plans to open the program to some taxpayers in all states and Washington, DC.

The Biden administration is touting the pilot program – which was funded by the Democrat-backed Inflation Reduction Act – as a win. The legislation, which passed in 2022, provided the IRS with a massive, 10-year investment meant to modernize taxpayer services and crack down on wealthy tax cheats. Republicans, concerned that small businesses and the middle class could be targeted by IRS auditors, have made several efforts to chip away at the agency’s funding.

…The pilot program has cost nearly $32m so far. About $75m has been budgeted for Direct File during fiscal year 2025, though the cost will depend on how many people ultimately use the program.

…A government survey found that 90% of Direct File users said their experience with the tax filing system was “excellent” or “above average.” Nearly half of the survey respondents had paid to file their federal tax returns the previous year.

«

This is a free replacement for the widely hated and paid-for Intuit Turbotax, which rakes in billions (not all from US personal tax filings). A rare and encouraging example of Americans leaning on government to replace a private software offering, and doing it better, creating real consumer surplus.
unique link to this extract

Google confirms the leaked Search documents are real • The Verge

Mia Sato:

»

A collection of 2,500 leaked internal documents from Google filled with details about data the company collects is authentic, the company confirmed on Wednesday. Until now, Google had refused to comment on the materials.

The documents in question detail data that Google is keeping track of, some of which may be used in its closely guarded search ranking algorithm. The documents offer an unprecedented — though still murky — look under the hood of one of the most consequential systems shaping the web.

“We would caution against making inaccurate assumptions about Search based on out-of-context, outdated, or incomplete information,” Google spokesperson Davis Thompson told The Verge in an email. “We’ve shared extensive information about how Search works and the types of factors that our systems weigh, while also working to protect the integrity of our results from manipulation.”

The existence of the leaked material was first outlined by search engine optimization (SEO) experts Rand Fishkin and Mike King, who each published initial analyses of the documents and their contents earlier this week. Google did not respond to The Verge’s multiple requests for comment yesterday about the authenticity of the leak.

The leaked material suggests that Google collects and potentially uses data that company representatives have said does not contribute to ranking webpages in Google Search, like clicks, Chrome user data, and more. The thousands of pages of documents act as a repository of information for Google employees, but it’s not clear what pieces of data detailed are actually used to rank search content — the information could be out of date, used strictly for training purposes, or collected but not used for Search specifically. The documents also do not reveal how different elements are weighted in search, if at all.

«

So why did Google take so long to confirm this? The documents have been around for days. One has to suspect that it was trying to run down the clock – had it confirmed them at once, that would make the story bigger in the wider tech journalism world – and to find a sufficiently minimal response. Notice how it doesn’t say that the information itself is out of context, outdated or incomplete. Just that if you ever found stuff that was, you should be wary.
unique link to this extract

Why publishers are preparing to federate their sites • Digiday

Sara Guaglione

»

At least two digital media companies are exploring the fediverse as a way to take more control over their referral traffic and onsite audience engagement. This comes at a time when walled gardens like Facebook and X are becoming less reliable for driving readers to publishers’ sites.

The Verge and 404 Media are building out new functions that would allow them to distribute posts on their sites and on federated platforms – like Threads, Mastodon and Bluesky – at the same time. Replies to those posts on those platforms become comments on their sites.

The fediverse allows users from different platforms and services to interact with one another without creating individual accounts for each platform, letting followers from one platform like and comment on a post on another platform. In other words, it lets social media networks that are independent from one another “talk” to each other.

But now publishers want in on the interoperability. The Verge, Vox Media’s tech news publication, and start-up tech site 404 Media are eager to federate their sites — they’re just waiting on the tech to support that functionality to become available to them.

“As an independent publisher, we are really excited about anything we can do to reach readers directly without needing to rely on social media platforms owned by massive tech companies who can take away access to our audience on a whim,” Jason Koebler, 404 Media co-founder, said in an email.

«

Have to love the lacuna between the headline, which promises a land grab, and the bathos of the opening phrase of “at least two”. But the principle is sound: Google and search engines which used to provide so much traffic are going to dry up, so get ahead of that shift.

However I think that if we’re talking about “social media platforms owned by massive tech companies”, Threads.. might fall into that category?
unique link to this extract

Nielsen launches the Media Distributor Gauge – the first convergent TV comparison of its kind • Nielsen

»

Across the primary viewing categories [in the US]:

• Cable was the only category in The Gauge to escape decline as it achieved a second consecutive monthly increase in share, moving up from 28.3% of TV in March to 29.1% (+0.8 pt.) in April. Cable sports viewing increased 28% vs. March, bolstered by NCAA basketball tournament coverage, NBA playoffs and the NFL draft. Women’s NCAA basketball finals and semifinals coverage accounted for four of the top six cable telecasts in April, and the WNBA draft notched 17th. While cable viewing increased about 1% on a monthly basis, a year-over-year comparison shows viewing has declined 8.2% vs. April 2023, and its share has lost 2.4 points.

• Broadcast viewing was down 3% in April, which equated to a 22.2% share of TV (-0.3 pt.). Similar to cable, women’s sports were the bright spot in the broadcast category this month. The NCAA women’s basketball championship game drew 17.6 million viewers on ABC (plus over 1 million more tuned in on ESPN), making it the top broadcast telecast in April by a large margin. The drama genre accounted for 29% of broadcast viewing, driven by Tracker, NCIS and Young Sheldon on CBS, and Chicago Fire and Chicago Med on NBC.

• Streaming viewership declined 1.9% from March to April, prompting the category to lose just 0.1 share point to account for 38.4% of total television. Amazon Prime Video saw the largest increase among streaming services this month with a 12% monthly increase for 3.2% of TV (+0.4 pt.). Prime Video’s April success was driven by its original series Fallout, which also topped all streaming titles this month with over 7 billion viewing minutes. YouTube, despite a 3% monthly decline in viewing, added a 15th month to its streak as the top streaming platform in The Gauge with a 9.6% share of TV in April.

«

There’s also a graphic of how time is spent watching what sort of TV. Note that YouTube TV is counted in Streaming – making up 10% of that category, or 3.8% overall – and comes ahead of Netflix. But for all the billions spent on streaming channels, they’re less than 40% of all viewing, and cable plus broadcast still accounts for over 50% of viewing time.

This is the US, of course. Other countries might be even more resilient against streaming.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

The British band Bring Me The Horizon included a hacking Easter egg in its new album. CC-licensed photo by April Woronowicz on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 10 links for you. Play the stream backwards. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

Delhi ‘unbearable’ as temperatures soar past 50ºC • BBC News

Meryl Sebastian and Kathryn Armstrong:

»

Parts of northern and central India are sweltering under a severe heatwave, with a provisional record temperature of 52.3ºC (126.1ºF) registered in Delhi. If verified, it would be the highest ever recorded in India.

More than 37 cities in the country recorded temperatures over 45ºC this week.

Warnings of heat-related illnesses have been issued, with at least three deaths reported so far.
The India Meteorological Department (IMD)’s Soma Sen Roy told the BBC that a team had been sent to the Mungeshpur area in Delhi – where the 52.3ºC temperature was recorded – to verify it.

The IMD described the recording as an “outlier compared to other stations”, which had recorded temperatures ranging from 45.2ºC to 49.1ºC in different parts of Delhi.

The city’s authorities have warned they will issue fines to those caught wasting water as the city deals with shortages and supplies have been cut to some areas.

Water minister Atishi announced that 200 teams would be deployed to crack down on people washing their cars with hosepipes and letting their tanks overflow.

“It’s been excruciatingly hot over the past couple of days and it’s got significantly worse as the days progress,” said BBC Business Correspondent Arunoday Mukharji, who is in Delhi.

A resident told news agency ANI earlier in the week that it was difficult to even eat properly because of the heat. “We have faced heat earlier as well, but this time it feels unbearable,” they said. “It’s difficult to even stand outside.”

The city’s power demand has soared to an all-time high, with residents turning to air conditioning, coolers and ceiling fans to cope with the heat.

A consumer court stopped hearing cases on Tuesday after the judge said it was too hot to work without air conditioning.

«

What isn’t mentioned is the “wet bulb” temperature. That’s what can decide literal life or death.
unique link to this extract

How it feels to get an AI email from a friend • Mrgan.com

Neven Morgan:

»

Recently I received an AI-written email from a friend. It wasn’t sent to test AI, or to show it off, as in “ha ha check this out”; my friend had a question to ask me, and the email asked it over the course of a few paragraphs. It then disclosed that, oh by the way, I used AI to write this. My reaction to this surprised me: I was repelled, as if digital anthrax had poured out of the app. I’m trying to figure out why.

…What this did feel like was, it felt as if my friend had buzzed their secretary over the intercom and barked at them to send me a letter, signed “R. Jeeves on behalf of ——————.”

• It felt like getting a birthday card with only the prewritten message inside, and no added well-wishes from the wisher’s own pen. An item off the shelf, paid for and handed over, transaction complete.

•Like getting a form letter, one of many thousands sent out by a large agency; except it was sent to me by, you know, a friend.

• It felt like the episode of Mrs. Maisel where Midge discovers that her husband’s comedy act features stolen Bob Newhart jokes.

• It felt like a family fridge decorated with printed stock art of children’s drawings.

• It felt like opening the front door at my birthday party to welcome in a group of iPads on wheels instead of people I like.

•It was like using a phone tree to kiss.

«

I think the friend needs to reconsider his/her priorities. (Almost certainly “his”, isn’t it.)
unique link to this extract

Once a sheriff’s deputy in Florida, now a source of disinformation from Russia • The New York Times

Steven Lee Myers:

»

A dozen years ago, John Mark Dougan, a former deputy sheriff in Palm Beach County, Fla., sent voters an email posing as a county commissioner, urging them to oppose the re-election of the county’s sheriff.

He later masqueraded online as a Russian tech worker with a pseudonym, BadVolf, to leak confidential information in violation of state law, fooling officials in Florida who thought they were dealing with a foreigner.

He also posed as a fictional New York City heiress he called Jessica, tricking an adviser to the Palm Beach County Sheriff’s Office into divulging improper conduct by the department. “And boy, did he ever spill ALL of the beans,” Mr. Dougan said in a written response to questions for this article, in which he confirmed his role in these episodes.

Those subterfuges in the United States, it turned out, were only a prelude to a more prominent and potentially more ominous campaign of deception he has been conducting from Russia.

Mr. Dougan, 51, who received political asylum in Moscow, is now a key player in Russia’s disinformation operations against the West. Back in 2016, when the Kremlin interfered in the American presidential election, an army of computer trolls toiled for hours in an office building in St. Petersburg to try to fool Americans online.

Today Mr. Dougan may be accomplishing much the same task largely by himself, according to American and European government officials and researchers from companies and organizations that have tracked his activities since August. The groups include NewsGuard, a company that reviews the reliability of news and information online; Recorded Future, a threat intelligence company; and Clemson University’s Media Forensics Hub.

Working from an apartment crowded with servers and other computer equipment, Mr. Dougan has built an ever-growing network of more than 160 fake websites that mimic news outlets in the United States, Britain and France.

With the help of commercially available artificial intelligence tools, including OpenAI’s ChatGPT and DALL-E 3, he has filled the sites with tens of thousands of articles, many based on actual news events. Interspersed among them are also bespoke fabrications that officials in the United States and European Union have attributed to Russian intelligence agencies or the administration of President Vladimir Putin.

Between September and May, Mr. Dougan’s outlets have been cited or referred to in news articles or social media posts nearly 8,000 times, and seen by more than 37 million people in 16 languages, according to a report released Wednesday by NewsGuard.

«

unique link to this extract

Vox Media and The Atlantic sign content deals with OpenAI • The Verge

Emilia David:

»

Two more media companies have signed licensing agreements with OpenAI, allowing their content to be used to train its AI models and be shared inside of ChatGPT. The Atlantic and Vox Media — The Verge’s parent company — both announced deals with OpenAI on Wednesday.

OpenAI has been quickly signing partnerships across the media world as it seeks to license training data and avoid copyright lawsuits. It’s recently reached deals with News Corp (The Wall Street Journal, the New York Post, and The Daily Telegraph), Axel Springer (Business Insider and Politico), DotDash Meredith (People, Better Homes & Gardens, Investopedia, Food & Wine, and InStyle), the Financial Times, and The Associated Press.

The deals appear to range in price based on the number of publications included. News Corp’s deal with OpenAI is estimated to be worth $250m over the next five years, according to the Journal, while the deal with the Financial Times is believed to be worth $5m to $10m. Terms for the deals with The Atlantic and Vox Media weren’t disclosed.

The agreements also cover how content from the publishers is displayed inside of ChatGPT. Content from Vox Media — including articles from The Verge, Vox, New York Magazine, Eater, SBNation, and their archives — and The Atlantic will get attribution links when it’s cited.

«

I didn’t expect ChatGPT to cite actual articles in its responses. It seems like the chatbot is turning into a search engine while Google, the search engine, turns itself into a chatbot.
unique link to this extract

Rock band’s hidden hacking-themed website gets hacked • TechCrunch

Lorenzo Franceschi-Bicchierai:

»

On Friday, Pal Kovacs was listening to the long-awaited new album from rock and metal giants Bring Me The Horizon when he noticed a strange sound at the end of the record’s last track.

Being a fan of solving riddles and breaking encrypted codes, Kovacs wondered, Does this sound contain a hidden message?

His hunch led to the discovery of a hidden hacking-themed website that at some point actually got hacked.
Kovacs opened the song in the audio-editing app Audacity and, as he suspected, there was indeed a spectrogram — essentially a visual representation of the audio itself — which was actually a scannable QR code. Excited, Kovacs shared his findings on the Bring Me The Horizon’s subreddit.

The QR code led to the hidden website, which is protected by a passcode that turned out to be a number (93934521) written on the album cover art, on the head of one of the characters, called M8. This M8 character speaks in some of the tracks and appears on the hidden site as a sort of guide.

The website is essentially an “alternate reality game,” or ARG, which bands like Nine Inch Nails have done before as a way to get fans more engaged with the band’s music and lore. 

In this particular case, the game consists of a website where, among other things, the band uploaded some unreleased tracks, a folder protected by a “cipher,” which led to more password-protected files, more mysteries, and more hidden Easter eggs, some of which are still unresolved and locked by unknown codes.

«

I’d never heard of Bring Me The Horizon: British band, apparently. It’s what I think of as cheeseburger rock – lots of layers which all get in the way of the principal element. (Style note: the band’s name has all initial capitals, but the Techcrunch stylistas downcapped “the” because that’s what they do in headlines. Bad form!)
unique link to this extract

Increasing use of renewable energy in US yields billions of dollars of benefits • The Guardian

Dharna Noor:

»

“From 2019 through 2022, wind and solar generation increased by about 55%,” said Dev Millstein, a scientist at the Lawrence Berkeley National Laboratory. “By 2022, wind and solar provided roughly 14% of total electricity needs for the US.”

During that time period, by reducing the use of fossil fuel power plants, the nation’s use of wind and solar power cut its carbon dioxide emissions by 900m metric tons, the authors found. That’s the equivalent of taking 71m cars off the road every year.

Those major climate benefits can obscure the air quality benefits renewable power yielded, wrote the authors, from the Lawrence Berkeley National Laboratory and the renewable consulting firm Clean Kilowatts. To illuminate those co-benefits, the researchers quantified how much the use of wind and solar reduced toxic air emissions, focusing specifically on sulfur dioxide (SO2) and nitrogen dioxides (NOx), which are both produced during fossil fuel combustion.

They found emissions of SO2 and NOx – both linked to increased asthma risk and a variety of other health issues – decreased by a total of 1m metric tons over that three-year period.

To determine the impact of that reduction on public health, the authors “used air quality models to track the population exposed to pollution from power plants”, Millstein said. They also employed epidemiological research to examine the effects of those emissions, and quantified the benefits by using an Environmental Protection Agency dollar value establishing the value of reducing the risk of early death across the population, he said.

All told, the emission reductions from SO2 and NOx provided $249bn of climate and health benefits to the US, the authors found – a figure Millstein said he found was “noteworthy”.

«

unique link to this extract

Meta takes down ‘inauthentic’ accounts on Facebook, Instagram linked to Israeli firm • WSJ

Salvador Rodriguez, Sam Schechner and Dustin Volz:

»

Meta Platforms said its security team took down a network of hundreds of fake accounts on Facebook and Instagram linked to an Israeli technology firm that apparently used AI-generated comments to praise Israel and criticize campus antisemitism. 

A network of more than 500 Facebook accounts and 32 Instagram accounts linked to an Israeli digital marketing and business-intelligence firm left comments on the pages of media organizations as well as political and public figures, including U.S. lawmakers, Meta said. The accounts posed as US and Canadian locals, including Jewish students and African-Americans, but were actually fake or compromised. 

Meta didn’t link the fake accounts to the Israeli government. Campaigns to spread false information on social media can also emanate from commercial actors or political campaigns, disinformation researchers say.   

The comments, which wouldn’t have been banned if they were from authentic users, were primarily in English and included calls for the release of Israeli hostages, praise for Israel’s military actions and criticisms of “radical Islam” and campus antisemitism. 

A large portion of the inauthentic network, which also had a presence on other social platforms, was detected and disabled by Meta’s automated systems before its investigation began, the company said in a report released Wednesday. The company said about 500 accounts followed the pages. Fewer than 100 accounts joined a Facebook group created by the inauthentic network, and about 2,000 accounts followed the Instagram profiles.

…Some of the comments referred to in Wednesday’s report were likely generated using artificial intelligence, Meta said, giving a glimpse of how digital activists and online campaigners are starting to mobilize the new technology to add scale to their efforts to spread false or inauthentic claims on social media.

…Many of the potentially AI-generated comments that Meta cited on Wednesday weren’t related to the posts to which they responded, a tactic Meta has seen in unsophisticated campaigns, the company said.

«

Notice the accounts were fake or compromised. One wonders what penalties Meta would impose on companies which use hacked accounts – though there might be the problem of proving that the company knew the accounts were hacked, or had done the hacking itself.
unique link to this extract

New BYD hybrid can drive non-stop for more than 2,000km (1,250 miles) • Bloomberg via Yahoo

Danny Lee:

»

BYD Co. unveiled a new hybrid powertrain capable of traveling more than 2,000 kilometers (1,250 miles) without recharging or refueling, intensifying the EV transition competition with the likes of Toyota and Volkswagen.

The upgraded tech, which aims to put more distance between BYD and its rivals, will be launched in two sedans immediately that cost under 100,000 yuan ($13,800), the automaker said at an event live-streamed Tuesday evening from China.

The longer range means some of BYD’s dual-mode plug-in electric hybrid cars can cover the equivalent of Singapore to Bangkok, New York to Miami, or Munich to Madrid on each charge and full tank of gas. The milestone marks BYD’s latest achievement in slashing fuel consumption since introducing its first hybrids in 2008.

Shenzhen-based BYD has upended China’s auto market with widespread price cuts — at some expense to its own profitability — and the positioning of the long-range hybrids may further stoke the price war. The company sold 3 million cars last year and has delivered almost 1 million this year through April. One of every two hybrids sold in China is a BYD, underlining the extent to which they’re a key revenue and profit driver for the manufacturer.

«

BYD is ahead of the western carmakers – it stopped making purely fossil fuel-powered ones in 2022 – and rather as with PCs and smartphones, China is showing that it can do as well or better and at greater scale than the west. And Biden’s looking to stop this?
unique link to this extract

Charges and seizures brought in fraud scheme, aimed at denying revenue for workers associated with North Korea • United States Department of Justice

»

The Justice Department unsealed charges, seizures, and other court-authorized actions to disrupt the illicit revenue generation efforts of the Democratic People’s Republic of Korea (DPRK or North Korea).

The charges include prosecutions of an Arizona woman, Ukrainian man, and three unidentified foreign nationals who allegedly participated in schemes to place overseas information technology (IT) workers—posing as US citizens and residents—in remote positions at U.S. companies. 

As alleged in the court documents, DPRK has dispatched thousands of skilled IT workers around the world, who used stolen or borrowed US persons’ identities to pose as domestic workers, infiltrate domestic companies’ networks, and raise revenue for North Korea. The schemes described in court documents involved defrauding over 300 US companies using US payment platforms and online job site accounts, proxy computers located in the US, and witting and unwitting US persons and entities. This announcement includes the largest case ever charged by the Justice Department involving this type of IT workers’ scheme.

…“As alleged in the indictment, Chapman and her co-conspirators committed fraud and stole the identities of American citizens to enable individuals based overseas to pose as domestic, remote IT workers,” said Principal Deputy Assistant Attorney General Nicole M. Argentieri, head of the Justice Department’s Criminal Division. “The charges in this case should be a wakeup call for American companies and government agencies that employ remote IT workers. These crimes benefitted the North Korean government, giving it a revenue stream and, in some instances, proprietary information stolen by the co-conspirators. The Criminal Division remains firm in its commitment to prosecute complex criminal schemes like this one.”

«

Endlessly inventive, the North Koreans. (This prosecution sponsored by The Owners Of Big City Office Blocks Where Workers Used To Work Every Damn Day.)
unique link to this extract

Internet Archive and the Wayback Machine under DDoS cyber-attack • Internet Archive Blogs

Chris Freeland:

»

The Internet Archive, the nonprofit research library that’s home to millions of historical documents, preserved websites, and media content, is currently in its third day of warding off an intermittent DDoS (distributed denial-of-service) cyber-attack. According to library staff, the collections are safe, though service remains inconsistent. Access to the Internet Archive Wayback Machine – which preserves the history of more than 866 billion web pages – has also been impacted.

Since the attacks began on Sunday, the DDoS intrusion has been launching tens of thousands of fake information requests per second. The source of the attack is unknown.

“Thankfully the collections are safe, but we are sorry that the denial-of-service attack has knocked us offline intermittently during these last three days,” explained Brewster Kahle, founder and digital librarian of the Internet Archive. “With the support from others and the hard work of staff we are hardening our defenses to provide more reliable access to our library. What is new is this attack has been sustained, impactful, targeted, adaptive, and importantly, mean.”

Cyber-attacks are increasingly frequent against libraries and other knowledge institutions, with the British Library, the Solano County Public Library (California), the Berlin Natural History Museum, and Ontario’s London Public Library all being recent victims.

«

The impossibility of discovering the source of DDOS attacks (by their nature) is always one of those big concerns. Why, too, would you carry out a DDOS against the Internet Archive? For fun? It can’t be for profit.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

iFixit and Samsung have broken off their device repair partnership after many stormy years. CC-licensed photo by spline splinson on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 10 links for you. Can’t fix this. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

Google won’t comment on a potentially massive leak of its search algorithm documentation • The Verge

Mia Sato:

»

an explosive leak that purports to show thousands of pages of internal documents appears to offer an unprecedented look under the hood of how Search works — and suggests that Google hasn’t been entirely truthful about it for years. So far, Google hasn’t responded to multiple requests for comment on the legitimacy of the documents.

Rand Fishkin, who worked in SEO for more than a decade, says a source shared 2,500 pages of documents with him with the hopes that reporting on the leak would counter the “lies” that Google employees had shared about how the search algorithm works. The documents outline Google’s search API and break down what information is available to employees, according to Fishkin.

The details shared by Fishkin are dense and technical, likely more legible to developers and SEO experts than the layperson. The contents of the leak are also not necessarily proof that Google uses the specific data and signals it mentions for search rankings. Rather, the leak outlines what data Google collects from webpages, sites, and searchers and offers indirect hints to SEO experts about what Google seems to care about, as SEO expert Mike King wrote in his overview of the documents.

The leaked documents touch on topics like what kind of data Google collects and uses, which sites Google elevates for sensitive topics like elections, how Google handles small websites, and more. Some information in the documents appears to be in conflict with public statements by Google representatives, according to Fishkin and King.

“‘Lied’ is harsh, but it’s the only accurate word to use here,” King writes. “While I don’t necessarily fault Google’s public representatives for protecting their proprietary information, I do take issue with their efforts to actively discredit people in the marketing, tech, and journalism worlds who have presented reproducible discoveries.”

«

Seems to have been exposed accidentally by Google when it was posted to Github and made public for a couple of months. It seems to be the biggest leak of documents ever from inside Google search. Hell of a thing: for all that the search algorithm has been kept incredibly secret for 30 years, the hacking was done by.. Google itself.
unique link to this extract

Google researchers say AI now leading disinformation vector (and are severely undercounting the problem) • 404 Media

Emanuel Maiberg:

»

As an endless stream of entirely wrong and sometimes dangerous AI-generated answers from Google are going viral on social media, new research from Google researchers and several fact checking organizations have found that most image-based disinformation is now AI-generated, but the way researchers collected their data suggests that the problem is even worse than they claim.

The paper, first spotted by the Faked Up newsletter, measures the rise of AI-generated image-based disinformation by looking at what fact checkers at Snopes, Politifact, and other sites have claimed were image-based disinformation. Overall, the study looks at a total of 135,838 fact checks which date back to 1995, but the majority of the claims were created after 2016 and the introduction of ClaimReview, a tagging system that allows fact checkers and publishers to flag disinformation for platforms like Google, Facebook, Bing, and others.

The most telling chart in the study shows the “prevalence of content manipulation types as a function of overall content manipulations.” In other words, it shows the different types of image-based disinformation and how common they are over time.  

As you can see from the chart, AI-generated image-based disinformation was just not a thing until late 2023, when AI image generators became widely available and popular, at which point they basically almost replaced all other forms of image-based disinformation. The chart also shows that there’s a slight increase in the total samples of image-based disinformation that corresponds with the rise of AI images, but only slightly. 

“Interestingly, the rise of AI images did not produce a bump in the overall proportion of misinformation claims that depend on images during this period, and image-based misinformation continued to decline on a relative basis as video-based misinformation grew,” the paper says.

«

The problem is in reality worse, because fact-checked content is only a small part of it.
unique link to this extract

Surveillance and interference: Israel’s covert war on the ICC exposed • +972 Magazine

Yuval Abraham and Meron Rapoport:

»

For nearly a decade, Israel has been surveilling senior International Criminal Court officials and Palestinian human rights workers as part of a secret operation to thwart the ICC’s probe into alleged war crimes, a joint investigation by +972 Magazine, Local Call, and the Guardian can reveal.

The multi-agency operation, which dates back to 2015, has seen Israel’s intelligence community routinely surveil the court’s current chief prosecutor Karim Khan, his predecessor Fatou Bensouda, and dozens of other ICC and UN officials. Israeli intelligence also monitored materials that the Palestinian Authority submitted to the prosecutor’s office, and surveilled employees at four Palestinian human rights organizations whose submissions are central to the probe.

According to sources, the covert operation mobilized the highest branches of Israel’s government, the intelligence community, and both the civilian and military legal systems in order to derail the probe.

The intelligence information obtained via surveillance was passed on to a secret team of top Israeli government lawyers and diplomats, who traveled to The Hague for confidential meetings with ICC officials in an attempt to “feed [the chief prosecutor] information that would make her doubt the basis of her right to be dealing with this question.” The intelligence was also used by the Israeli military to retroactively open investigations into incidents that were of interest to the ICC, to try to prove that Israel’s legal system is capable of holding its own to account.

Additionally, as the Guardian reported earlier today, the Mossad, Israel’s foreign intelligence agency, ran its own parallel operation which sought out compromising information on Bensouda and her close family members in an apparent attempt to sabotage the ICC’s investigation. The agency’s former head, Yossi Cohen, personally attempted to “enlist” Bensouda and manipulate her into complying with Israel’s wishes, according to sources familiar with his activities, causing the then-prosecutor to fear for her personal safety.

«

Hell of an operation for a court that Israel feigns not to be concerned about. Also, perhaps we should simply assume that all operations are being surveilled all the time.
unique link to this extract

GPT-4o’s Chinese token-training data is polluted by spam and porn websites • MIT Technology Review

Zeyi Yang:

»

Soon after OpenAI released GPT-4o on Monday, May 13, some Chinese speakers started to notice that something seemed off about this newest version of the chatbot: the tokens it uses to parse text were full of spam and porn phrases.

On May 14, Tianle Cai, a PhD student at Princeton University studying inference efficiency in large language models like those that power such chatbots, accessed GPT-4o’s public token library and pulled a list of the 100 longest Chinese tokens the model uses to parse and compress Chinese prompts. 

Humans read in words, but LLMs read in tokens, which are distinct units in a sentence that have consistent and significant meanings. Besides dictionary words, they also include suffixes, common expressions, names, and more. The more tokens a model encodes, the faster the model can “read” a sentence and the less computing power it consumes, thus making the response cheaper.

Of the 100 results, only three of them are common enough to be used in everyday conversations; everything else consisted of words and expressions used specifically in the contexts of either gambling or pornography. The longest token, lasting 10.5 Chinese characters, literally means “_free Japanese porn video to watch.” Oops.

«

This was so, so predictable.
unique link to this extract

Analysis: monthly drop hints that China’s CO2 emissions may have peaked in 2023 • Carbon Brief

Lauri Myllyvirta:

»

China’s carbon dioxide (CO2) emissions fell by 3% in March 2024, ending a 14-month surge that began when the economy reopened after the nation’s “zero-Covid” controls were lifted in December 2022.

The new analysis for Carbon Brief, based on official figures and commercial data, reinforces the view that China’s emissions could have peaked in 2023.

The drivers of the CO2 drop in March 2024 were expanding solar and wind generation, which covered 90% of the growth in electricity demand, as well as declining construction activity.

Oil demand growth also ground to a halt, indicating that the post-Covid rebound may have run its course.

A 2023 peak in China’s CO2 emissions is possible if the buildout of clean energy sources is kept at the record levels seen last year.

However, there are divergent views across the industry and government on the outlook for clean energy growth. How this gap gets resolved is the key determinant of when China’s emissions will peak – if they have not done so already.

Other key findings from the analysis include:
• Wind and solar growth pushed fossil fuels’ share of electricity generation in China down to 63.6% in March 2024, from 67.4% a year earlier, despite strong growth in demand
• The ongoing contraction of real-estate construction activity in China saw steel production fall by 8% and cement output by 22% in March 2024
• Electric vehicles (EVs) now make up around one-in-10 vehicles on China’s roads, knocking around 3.5 percentage points off the growth in petrol demand
• Some 45% of last year’s record solar additions were smaller-scale “distributed” systems, creating an illusory “missing data problem”. 

«

So fossil fuels are still a very, very big source of generation – lots of it coal – but the picture keeps improving. And the EV stat helps too.
unique link to this extract

What does the public in six countries think of generative AI in news? • Reuters Institute for the Study of Journalism

Dr Richard Fletcher and Professor Rasmus Kleis Nielsen:

»

Based on an online survey focused on understanding if and how people use generative artificial intelligence (AI), and what they think about its application in journalism and other areas of work and life across six countries (Argentina, Denmark, France, Japan, the UK, and the USA), we present the following findings.

Findings on the public’s use of generative AI: ChatGPT is by far the most widely recognised generative AI product – around 50% of the online population in the six countries surveyed have heard of it. It is also by far the most widely used generative AI tool in the six countries surveyed. That being said, frequent use of ChatGPT is rare, with just 1% using it on a daily basis in Japan, rising to 2% in France and the UK, and 7% in the USA. Many of those who say they have used generative AI have used it just once or twice, and it is yet to become part of people’s routine internet use.

In more detail, we find:
• While there is widespread awareness of generative AI overall, a sizable minority of the public – between 20% and 30% of the online population in the six countries surveyed – have not heard of any of the most popular AI tools
• In terms of use, ChatGPT is by far the most widely used generative AI tool in the six countries surveyed, two or three times more widespread than the next most widely used products, Google Gemini and Microsoft Copilot
• Younger people are much more likely to use generative AI products on a regular basis. Averaging across all six countries, 56% of 18–24s say they have used ChatGPT at least once, compared to 16% of those aged 55 and over
• Roughly equal proportions across six countries say that they have used generative AI for getting information (24%) as creating various kinds of media, including text but also audio, code, images, and video (28%)
• Just 5% across the six countries covered say that they have used generative AI to get the latest news.

«

That last 5% worries me, since generative AI has no business offering news. Notable too how it doesn’t seem to stick: people try it out and give it up.
unique link to this extract

iFixit ends Samsung deal as oppressive repair shop requirements come to light • Ars Technica

Ron Amadeo:

»

IFixit and Samsung were once leading the charge in device repair, but iFixit says it’s ending its repair partnership with Samsung because it feels Samsung just isn’t participating in good faith. iFixit says the two companies “have not been able to deliver” on the promise of a viable repair ecosystem, so it would rather shut the project down than continue. The repair site says “flashy press releases and ambitious initiatives don’t mean much without follow-through.”

iFixit’s Scott Head explains: “As we tried to build this ecosystem we consistently faced obstacles that made us doubt Samsung’s commitment to making repair more accessible. We couldn’t get parts to local repair shops at prices and quantities that made business sense. The part prices were so costly that many consumers opted to replace their devices rather than repair them. And the design of Samsung’s Galaxy devices remained frustratingly glued together, forcing us to sell batteries and screens in pre-glued bundles that increased the cost.”

…Samsung has also reportedly been on the attack against repair, even while it partners with iFixit. On the same day that iFixit announced it was dropping the partnership, 404 Media reported that Samsung requires independent repair shops to turn over customer data and “immediately disassemble” any device found to be using third-party parts. Imagine taking your phone to a shop for repair and finding out it was destroyed by the shop as a requirement from Samsung. The report also says Samsung’s contracts require that independent companies “daily” upload to a Samsung database (called G-SPN) the details of each and every repair “at the time of each repair.”

«

(Thanks G for the link.)
unique link to this extract

Apple wants to know if you’re hearing things because of tinnitus • The Verge

Justine Calma:

»

More than 77% of people who participated in a big Apple-sponsored study have experienced tinnitus at some point in their lives, according to preliminary data. Around 15% say they’re affected daily by tinnitus, perceiving ringing or other sounds that other people can’t hear.

In one of the largest surveys of its kind, researchers at the University of Michigan gathered data from more than 160,000 participants who responded to survey questions and completed hearing assessments on Apple’s Research app since 2019. The goal is to study the effects of sound exposure through headphones, how tinnitus impacts people, and perhaps develop new methods for managing the symptoms.

“The trends that we’re learning through the Apple Hearing Study about people’s experience with tinnitus can help us better understand the groups most at risk, which can in turn help guide efforts to reduce the impacts associated with it,” University of Michigan environmental health sciences professor Rick Neitzel said in a press release.

«

unique link to this extract

Forget retirement. Older people are turning to gig work to survive • Rest of World

Laís Martins, Kimberly Mutandiro, Lam Le and Zuha Siddiqui:

»

Most gig workers around the world are relatively young: Research published in 2021 by the International Labour Organization (ILO), a United Nations agency focused on improving working conditions, puts the average age for delivery workers at 29 and the average age for ride-hailing drivers at 36. But Rest of World reporting suggests that older individuals are turning to gig work, too — and their numbers are expected to grow in the coming years.

Over the past three months, Rest of World spoke to 52 gig workers between the ages of 50 and 75 years in Latin America, Africa, South Asia, and Southeast Asia. Some chose gig work to keep up with rising living costs or to make up for threadbare social security systems; others say it’s impossible to find employment elsewhere once they near retirement age. Still others say that gig work is a second job, one they can transition into full-time once they retire or are no longer employed. Many reported low earnings, long hours, and health complications from their work. Without being able to put enough savings aside for retirement, others communicated the feeling that there was no alternative. But for all its challenges and pitfalls, gig work can represent an accessible, flexible, low-barrier activity that enables older workers to offset expenses and continue to be active. 

… Older gig workers is a demographic that’s expected to grow in the coming decades. The global population of people 65 or older is expected to double by 2050, surpassing 1.6 billion, according to the U.N. At the same time, family units around the world are transforming, often requiring older people to support themselves for longer.

«

unique link to this extract

How researchers cracked an 11-year-old password to a $3m crypto wallet • WIRED

Condé Nast:

»

Two years ago when “Michael,” an owner of cryptocurrency, contacted Joe Grand to help recover access to about $2m worth of bitcoin he stored in encrypted format on his computer, Grand turned him down.

Michael, who is based in Europe and asked to remain anonymous, stored the cryptocurrency in a password-protected digital wallet. He generated a password using the RoboForm password manager and stored that password in a file encrypted with a tool called TrueCrypt. At some point, that file got corrupted and Michael lost access to the 20-character password he had generated to secure his 43.6 BTC (worth a total of about €4,000, or $5,300, in 2013). Michael used the RoboForm password manager to generate the password but did not store it in his manager. He worried that someone would hack his computer and obtain the password.

“At [that] time, I was really paranoid with my security,” he laughs.

…last June he approached Grand again, hoping to convince him to help, and this time Grand agreed to give it a try, working with a friend named Bruno in Germany who also hacks digital wallets.

Grand and Bruno spent months reverse engineering the version of the RoboForm program that they thought Michael had used in 2013 and found that the pseudo-random number generator used to generate passwords in that version—and subsequent versions until 2015—did indeed have a significant flaw that made the random number generator not so random. The RoboForm program unwisely tied the random passwords it generated to the date and time on the user’s computer—it determined the computer’s date and time, and then generated passwords that were predictable. If you knew the date and time and other parameters, you could compute any password that would have been generated on a certain date and time in the past.

If Michael knew the day or general time frame in 2013 when he generated it, as well as the parameters he used to generate the password (for example, the number of characters in the password, including lower- and upper-case letters, figures, and special characters), this would narrow the possible password guesses to a manageable number. Then they could hijack the RoboForm function responsible for checking the date and time on a computer and get it to travel back in time, believing the current date was a day in the 2013 time frame when Michael generated his password. RoboForm would then spit out the same passwords it generated on the days in 2013.

There was one problem: Michael couldn’t remember when he created the password.

«

So, so many tales of bitcoin people who lose their password.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

Zebrafish are surviving, so far, on the Chinese space station – but seem a little disoriented. CC-licensed photo by Oregon State University on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 10 links for you. Which was is up, though? I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

How China’s 1980s PC industry hacked dot-matrix printers • Fast Company

Thomas Mullaney:

»

Commercial dot-matrix printing was yet another arena in which the needs of Chinese character I/O were not accounted for. This is witnessed most clearly in the then-dominant configuration of printer heads—specifically the 9-pin printer heads found in mass-manufactured dot-matrix printers during the 1970s. Using nine pins, these early dot-matrix printers were able to produce low-resolution Latin alphabet bitmaps with just one pass of the printer head. The choice of nine pins, in other words, was “tuned” to the needs of Latin alphabetic script.

These same printer heads were incapable of printing low-resolution Chinese character bitmaps using anything less than two full passes of the printer head, one below the other. Two-pass printing dramatically increased the time needed to print Chinese as compared to English, however, and introduced graphical inaccuracies, whether due to inconsistencies in the advancement of the platen or uneven ink registration (that is, characters with differing ink densities on their upper and lower halves).

Compounding these problems, Chinese characters printed in this way were twice the height of English words. This created comically distorted printouts in which English words appeared austere and economical, while Chinese characters appeared grotesquely oversized. Not only did this waste paper, but it left Chinese-language documents looking something like large-print children’s books. When consumers in the Chinese-Japanese-Korean (CJK) world began to import Western-manufactured dot-matrix printers, then, they faced yet another facet of Latin alphabetic bias.

«

This is an extract from what looks like a fascinating book about how China yanked itself into the computer age. Necessity as the mother of invention, and all that.
unique link to this extract

Fish are adapting to weightlessness on the Chinese space station • Universe Today

Scott Alan Johnston:

»

Four zebrafish are alive and well after nearly a month in space aboard China’s Tiangong space station. As part of an experiment testing the development of vertebrates in microgravity, the fish live and swim within a small habitat aboard the station.

While the zebrafish have thus far survived, they are showing some signs of disorientation. The taikonauts aboard Tiangong – Ye Guangfu, Li Cong, and Li Guangsu – have reported instances of swimming upside down, backward, and in circular motions, suggesting that microgravity is having an effect on their spatial awareness.

The zebrafish were launched aboard Shenzhou-18, which carried them, as well as a batch of hornwort, to orbit on April 25, 2024. The aim of the project is to create a self-sustaining ecosystem, studying the effects of both microgravity and radiation on the development and growth of these species.

As a test subject, zebrafish have several advantages. Their short reproductive and development cycle, and transparent eggs, allow scientists to study their growth quickly and effectively, and their genetic makeup shares similarities with humans, potentially offering insights that are relevant to human health. The zebrafish genome has been fully sequenced, and for these reasons zebrafish are commonly used in scientific experiments on Earth. Seeing how these well-studied creatures behave in such an extreme environment may have a lot to tell us about the life and development of vertebrates across species while exposed to microgravity.

«

“Showing some signs of disorientation” sounds like they’re doing remarkably well, all things considered. (My understanding is they were taken up as fertilised embryos, rather than adult fish, because I’d have questions about how fish experience – or survive – high G forces.)
unique link to this extract

AI headphones let wearer listen to a single person in a crowd, by looking at them just once • UW News

Stefan Milne and Kiyomi Taguchi:

»

A University of Washington team has developed an artificial intelligence system that lets a user wearing headphones look at a person speaking for three to five seconds to “enroll” them. The system, called “Target Speech Hearing,” then cancels all other sounds in the environment and plays just the enrolled speaker’s voice in real time even as the listener moves around in noisy places and no longer faces the speaker.

The team presented its findings May 14 in Honolulu at the ACM CHI Conference on Human Factors in Computing Systems. The code for the proof-of-concept device is available for others to build on. The system is not commercially available.

“We tend to think of AI now as web-based chatbots that answer questions,” said senior author Shyam Gollakota, a UW professor in the Paul G. Allen School of Computer Science & Engineering. “But in this project, we develop AI to modify the auditory perception of anyone wearing headphones, given their preferences. With our devices you can now hear a single speaker clearly even if you are in a noisy environment with lots of other people talking.”

To use the system, a person wearing off-the-shelf headphones fitted with microphones taps a button while directing their head at someone talking. The sound waves from that speaker’s voice then should reach the microphones on both sides of the headset simultaneously; there’s a 16º margin of error. The headphones send that signal to an on-board embedded computer, where the team’s machine learning software learns the desired speaker’s vocal patterns.

«

Nifty. Wonder how long it will take for this to be incorporated into commercial systems. (It would be madness to try to do this as a standalone commercial project. One hopes the UW team realise that.)
unique link to this extract

You were promised a jetpack by liars • Pluralistic

Cory Doctorow:

»

As a science fiction writer, I find it weird that some sf tropes – like space colonization – have become culture-war touchstones. You know, that whole “we were promised jetpacks” thing.

I confess, I never looked too hard at the practicalities of jetpacks, because they are so obviously either used as a visual shorthand (as in the Jetsons) or as a metaphor. Even a brief moment’s serious consideration should make it clear why we wouldn’t want the distracted, stoned, drunk, suicidal, homicidal maniacs who pilot their two-ton killbots through our residential streets at 75mph to be flying over our heads with a reservoir of high explosives strapped to their backs.

Jetpacks can make for interesting sf eyeball kicks or literary symbols, but I don’t actually want to live in a world of jetpacks. I just want to read about them, and, of course, write about them.

I had blithely assumed that this was the principle reason we never got the jetpacks we were “promised.” I mean, there kind of was a promise, right? I grew up seeing videos of rocketeers flying their jetpacks high above the heads of amazed crowds, at World’s Fairs and Disneyland and big public spectacles. There was that scene in Thunderball where James Bond (the canonical Connery Bond, no less) makes an escape by jetpack. There was even a Gilligan’s Island episode where the castaways find a jetpack and scheme to fly it all the way back to Hawai’i.

Clearly, jetpacks were possible, but they didn’t make any sense, so we decided not to use them, right?

Well, I was wrong. In a terrific new 99% Invisible episode, Chris Berube tracks the history of all those jetpacks we saw on TV for decades, and reveals that they were all the same jetpack, flown by just one guy, who risked his life every time he went up in it…

«

Terrifying. (Rather like Doctorow’s work rate. How in the world does he do it.)
unique link to this extract

A brief history of the traffic light and why we need a new colour • Al Jazeera

»

The universally known traffic light has not experienced a significant redesign in almost 100 years, ever since William Pott, a Detroit police officer, created the first three-section traffic light in the United States in 1921. Now, say experts, the rise of driverless cars means that a new set of safety guidelines is needed to ensure they interact correctly with traffic signals.

Traffic lights around the world typically use red, amber and green lights to signal to drivers whether they should stop, go or get ready to either stop or go at intersections and pedestrian crossings. Ali Hajbabaie, a North Carolina State University (NCSU) engineering professor, is leading a team to design a traffic system that considers how driverless cars respond to traffic signals.

Hajbabaie told The Associated Press news agency that he proposes adding another light – possibly a white one.

…Humans and autonomous cars use different sets of visual cues when it comes to interpreting lighting systems. Different colours – sometimes flashing to indicate that a change is imminent – work best for the human brain, while a single light works better for autonomous cars.

Therefore, a fourth light – most likely white – would be added for the benefit of self-driving cars. The white light would be interpreted by a self-driven car as an instruction to “keep going unless instructed otherwise”.

Hajbabaie, the NCSU professor, explained: “If the white light is active, you just follow the vehicle in front of you.”

«

This is going to make those captchas with “click the squares with traffic lights” much trickier. Also, though it sounds like a Beatles lyric, how many traffic lights would you have to change around the world? I’m going to guess hundreds of millions in existence.
unique link to this extract

Announcing Exa, a search engine built for the AI world

exa:

»

The internet contains the collective knowledge output of mankind – all the great works of art and literature, millions of essays, hundreds of millions of research papers, billions of images and videos, trillions of ideas sprinkled across tweets, forums, and memes.

Searching the internet should feel like navigating a grand library of knowledge, where you could weave insights across cultures, industries, and millenia.

Of course it doesn’t feel that way. Today, searching the internet feels more like navigating a landfill.

Many have debated what’s wrong with search.

But the core problem is actually simple – knowledge on the internet is buried under an overwhelming amount of information.

The core solution is also simple – we need a better search algorithm to filter all that information and organize the knowledge buried inside.

Exa is going to organize the world’s knowledge. Exa’s goal is to understand any query – no matter how complex – and filter the internet to exactly the knowledge required for that query.

To illustrate the problem, try googling “startups working on climate change”.

«

Interesting: I’ve noticed a few people saying “this is the new Google, like Google was back in 1999”. Notable that it needs you to ask in a slightly stilted style:

»

Exa uses a transformer architecture to predict links given text, and it gets its power from having been trained on the way that people talk about links on the Internet. This training produces a model that returns links that are both high in relevance and quality. However, the model does expect queries that look like how people describe a link on the Internet. For example, ‘best restaurants in SF” is a bad query, whereas “Here is the best restaurant in SF:” is a good query.

«

I tried it with a query local to me. It’s… a work in progress.
unique link to this extract

From pollution to solution • Cremieux Recueil

Andrew Song:

»

This post examines the transformative role that stratospheric aerosol injection of SO₂ can play in moderating change in Earth’s climate. While many fear the corrosive effects of SO₂ based on its familiar ground-level impacts, its application in the upper atmosphere tells a different story altogether—one of cooling the Earth effectively and potentially reversing some of the most severe effects of global warming.

…First, we’ll determine the total area affected. To do this, we’ll assume aerosols spread uniformly over the Earth’s surface, an area of approximately 510 million square kilometers.

Second, we’ll compute the sulfuric acid required to produce mildly acidic rain. To impact our designated area, we’ll need 0.01 metric tons of sulfuric acid per square kilometer. Thus, the total required sulfuric acid is 0.01 metric tons per square kilometer * 510 million square kilometers = 5.1 million metric tons.

Third, we’ll compute the SO₂ needed to produce mildly acidic rain. Considering the conversion efficiency of 70%, the total SO₂ needed would be 5.1 million metric tons over the 70% efficiency, or approximately 7.29 million metric tons of SO₂.

Thus, we have our upper limit of 7.29 million metric tons of SO₂ that can be pumped into the atmosphere safely each year. Any more and we’ll end up with acidic rain. The next question is, does the amount bring global temperatures down meaningfully? Luckily, this exact question has been modeled extensively by scientists and featured in a report by the United Nations Environment Programme, which states:

»

It is estimated that continuous injection rates of 8–16 million metric tons of sulphur dioxide (SO₂) per year (approximately equivalent to the estimated injection amount of Mount Pinatubo in the single year of 1991) would reduce global mean temperature by 1°C.

«

«

Song is a co-founder of Make Sunsets, “the first company in the world to commercialise stratospheric aerosol injection as a service to cool the Earth”. It has some big-name VC backers. The bigger question is whether we want private companies noodling with the atmosphere. (Though in a sense, they have been for centuries already.)
unique link to this extract

“I was misidentified as shoplifter by facial recognition tech” • BBC News

James Clayton:

»

Sara needed some chocolate – she had had one of those days – so wandered into a Home Bargains store.

“Within less than a minute, I’m approached by a store worker who comes up to me and says, ‘You’re a thief, you need to leave the store’.”

Sara – who wants to remain anonymous – was wrongly accused after being flagged by a facial-recognition system called Facewatch. She says after her bag was searched she was led out of the shop, and told she was banned from all stores using the technology.

“I was just crying and crying the entire journey home… I thought, ‘Oh, will my life be the same? I’m going to be looked at as a shoplifter when I’ve never stolen’.” Facewatch later wrote to Sara and acknowledged it had made an error.

Facewatch is used in numerous stores in the UK – including Budgens, Sports Direct and Costcutter – to identify shoplifters. The company declined to comment on Sara’s case to the BBC, but did say its technology helped to prevent crime and protect frontline workers. Home Bargains, too, declined to comment.

It’s not just retailers who are turning to the technology. On a humid day in Bethnal Green, in east London, we joined the police as they positioned a modified white van on the high street. Cameras attached to its roof captured thousands of images of people’s faces. If they matched people on a police watchlist, officers would speak to them and potentially arrest them.

Unflattering references to the technology liken the process to a supermarket checkout – where your face becomes a bar code.

On the day we were filming, the Metropolitan Police said they made six arrests with the assistance of the tech. That included two people who breached the terms of their sexual-harm prevention orders, a man wanted for grievous bodily harm and a person wanted for the assault of a police officer.

«

unique link to this extract

How BBC’s breaking news alerts are giving voters – and political parties – an electoral buzz • The Guardian

Jim Waterson:

»

he most powerful person in British media during this election, in terms of having the most direct access to voters, is no longer the editor of BBC’s News at Six or the person who chooses the headlines on Radio 2. Nor are they a newspaper editor, a TikTok influencer, or a podcaster. Instead, they’re the anonymous on-shift editor of the BBC News app, making snap judgments on whether to make the phones of millions of Britons buzz with a breaking news push alert.

The BBC does not publish user numbers, but external research suggests about 12.6 million Britons have its news app installed. BBC newsroom sources say the actual number is higher and the assumption is that about 60% of users have notifications enabled. This means that on a conservative estimate, a typical push alert is reaching the phones of seven million Britons – more than any other broadcast news bulletin in the UK.

Craig Oliver, David Cameron’s former director of communications, said influencing the BBC’s coverage was the main objective for all political press officers. This used to mean phoning the editors of specific television news shows. Now the focus is shifting online: “The sheer scale of the website alone and its breaking news alerts is huge. Once something gets into the water supply of the BBC, it’s very hard to get it out.”

The audience for the BBC’s News at Six, the most-watched television news show in the country, is down to about 3.5 million viewers – and its audience is ageing; the average BBC One television viewer is now in their 60s. Print newspaper front pages remain heavily analysed, but their reach has collapsed. Twenty years ago, the Sun still sold 3.5m copies a day. Now the biggest selling print newspaper is the Daily Mail, on 700,000 copies.

All have been quietly overtaken by the the BBC News app push alert, which was only widely adopted a decade ago. An alert can drive readers to open the full news story – or its headline can exist as a standalone nugget of news, a 25-word summary destined to remain unclicked. Some people are driven to distraction by breaking news buzzes, while others grab their phones the moment they see them arrive.

«

Waterson is billed as the “political media editor”, which I hope is only temporary during the election.
unique link to this extract

Google is playing a dangerous game with AI search • The Atlantic

Caroline Mimbs Nyce tried Google’s new AI search on health questions:

»

This risk with generative AI isn’t just about Google spitting out blatantly wrong, eyeroll-worthy answers. As the AI research scientist Margaret Mitchell tweeted, “This isn’t about ‘gotchas,’ this is about pointing out clearly foreseeable harms.” Most people, I hope, should know not to eat rocks. The bigger concern is smaller sourcing and reasoning errors—especially when someone is Googling for an immediate answer, and might be more likely to read nothing more than the AI overview. For instance, it told me that pregnant women could eat sushi as long as it doesn’t contain raw fish. Which is technically true, but basically all sushi has raw fish. When I asked about ADHD, it cited AccreditedSchoolsOnline.org, an irrelevant website about school quality.

When I Googled How effective is chemotherapy?, the AI overview said that the one-year survival rate is 52%. That statistic comes from a real scientific paper, but it’s specifically about head and neck cancers, and the survival rate for patients not receiving chemotherapy was far lower. The AI overview confidently bolded and highlighted the stat as if it applied to all cancers.

In certain instances, a search bot might genuinely be helpful. Wading through a huge list of Google search results can be a pain, especially compared with a chatbot response that sums it up for you. The tool might also get better with time.

Still, it may never be perfect. At Google’s size, content moderation is incredibly challenging even without generative AI. One Google executive told me last year that 15% of daily searches are ones the company has never seen before. Now Google Search is stuck with the same problems that other chatbots have: Companies can create rules about what they should and shouldn’t respond to, but they can’t always be enforced with precision.

«

Not it “may” never be perfect; it can never be perfect. Search is already incredibly difficult, and the sooner people – including people at Google – understand that the better.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

A number of hotels in the Wyndham chain turn out to have had spyware on their checkin computers – making guest details visible to hackers. CC-licensed photo by Greg Grimes on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

A selection of 10 links for you. Which credit card? I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

Google scrambles to manually remove weird AI answers in search • The Verge

Kylie Robison:

»

Social media is abuzz with examples of Google’s new AI Overview product saying weird stuff, from telling users to put glue on their pizza to suggesting they eat rocks. The messy rollout means Google is racing to manually disable AI Overviews for specific searches as various memes get posted, which is why users are seeing so many of them disappear shortly after being posted to social networks.

It’s an odd situation, since Google has been testing AI Overviews for a year now — the feature launched in beta in May 2023 as the Search Generative Experience — and CEO Sundar Pichai has said the company served over a billion queries in that time.

But Pichai has also said that Google’s brought the cost of delivering AI answers down by 80% over that same time, “driven by hardware, engineering and technical breakthroughs.” It appears that kind of optimization might have happened too early, before the tech was ready.

“A company once known for being at the cutting edge and shipping high-quality stuff is now known for low-quality output that’s getting meme’d,” one AI founder, who wished to remain anonymous, told The Verge.

Google continues to say that its AI Overview product largely outputs “high quality information” to users. “Many of the examples we’ve seen have been uncommon queries, and we’ve also seen examples that were doctored or that we couldn’t reproduce,” Google spokesperson Meghann Farnsworth said in an email to The Verge. Farnsworth also confirmed that the company is “taking swift action” to remove AI Overviews on certain queries “where appropriate under our content policies, and using these examples to develop broader improvements to our systems, some of which have already started to roll out.”

Gary Marcus, an AI expert and an emeritus professor of neural science at New York University, told The Verge that a lot of AI companies are “selling dreams” that this tech will go from 80% correct to 100%. Achieving the initial 80% is relatively straightforward since it involves approximating a large amount of human data, Marcus said, but the final 20% is extremely challenging. In fact, Marcus thinks that last 20% might be the hardest thing of all.

«

Two points: note that Google doesn’t say how the cost of delivering AI answers compares to normal searches; and every day 15% of queries are completely novel, which amounts to millions every day. Google will never succeed at this.
unique link to this extract

The Elmer’s Glue pizza error is more fascinating than you think • The End(s) of Argument

Mike Caulfield:

»

Since intent can’t always be accurately read, a good search result for an ambiguous query will often be a bit of a buffet, pulling a mix of search results from different topical domains. When I say, for example, “why do people like pancakes but hate waffles” I’m going to get two sorts of results. First, I’ll get a long string of conversations where the internet debates the virtues of pancakes vs. waffles. Second, I’ll get links and discussion about a famous Twitter post about the the hopelessness of conversation on Twitter.

For an ambiguous query, this is a good result set. If you see a bit of each in the first 10 blue links you can choose your own adventure here. It’s hard for Google to know what matches your intent, but it’s trivially easy for you to spot things that match your intent. In fact, the wider apart the nature of the items, the easier it is to spot the context that applies to you. So a good result set will have a majority of results for what you’re probably asking and a couple results for things you might be asking, and you get to choose your path.

Likewise, when you put in terms about cheese sliding off pizza, Google could restrict the returned results to recipe sites, advice which would be relatively glue-free. But maybe you want advice, or maybe you want to see discussion. Maybe you want to see jokes. Maybe you are looking for a very specific post you read, and not looking to solve an issue at all, in which case you just want relevance completely determined by closeness of word match. Maybe you’re looking for a movie scene you remember about cheese sliding off pizza.

In the beforetimes, it would be hard to imagine a user getting upset that the cheese sliding query pulls up a joke on Reddit as well as some recipe tips. The user can spot, quite easily, that these are two different sorts of things.

The problem comes when the results get synthesized into a common answer. To some extent, this is a form of “context collapse”, where the different use contexts (jokes, movies, recipes, whatever) get blended into a single context.

«

unique link to this extract

Things the guys who have my stolen phone have texted me to try to get me to unlock it • Read Max

Veronica de Souza:

»

My phone was stolen in early March, most likely while I was standing on the platform at the York Street station waiting for the F train. Fifteen minutes later, in a food hall under the movie theater where my boyfriend and I were supposed to see Dune 2, I reached into my pocket and realized it was gone. Josh looked at the “Find My” app on his phone: My phone was “last seen” at York Street but wasn’t registering a current location. Someone had turned it off.

As quickly as possible, I did all the things you’re supposed to do when your phone is lost or stolen–mark it as lost, cut off service, and remotely erase it–and spent the rest of the night anxiously refreshing the Find My App, watching my phone move around Manhattan before finally coming to a stop at Rockefeller Center. I didn’t bother confronting the thief.

Worst of all, we didn’t even see Dune 2.

After two hours in the Williamsburg Apple Store the next morning, I had a new iPhone 15 and I stopped stressing. As long as I didn’t remove the phone from my Apple account or the “Find My” app, the phone was essentially bricked to anyone without the passcode–unusable by the thieves, or the fences who I assume bought it from them. Now my phone was their problem.

…Here are some things the guys who stole (or later purchased) my iPhone have told me to try to get me to unlock it:

1) “Your iPhone 14 Pro is trying to pay with Apple pay in China.”

«

And that’s only the beginning. The fight between scammers and intelligent people who recognise scammers is always interesting.
unique link to this extract

Spyware found on US hotel check-in computers • TechCrunch

Zack Whittaker:

»

A consumer-grade spyware app has been found running on the check-in systems of at least three Wyndham hotels across the United States, TechCrunch has learned.

The app, called pcTattletale, stealthily and continually captured screenshots of the hotel booking systems, which contained guest details and customer information. Thanks to a security flaw in the spyware, these screenshots are available to anyone on the internet, not just the spyware’s intended users. 

This is the most recent example of consumer-grade spyware exposing sensitive information because of a security flaw in the spyware itself. It’s also the second known time that pcTattletale has exposed screenshots of the devices on which the app is installed. Several other spyware apps in recent years had security bugs or misconfigurations that exposed the private and personal data of unwitting device owners, in some cases prompting action by government regulators.

pcTattletale allows whomever controls it to remotely view the target’s Android or Windows device and its data, from anywhere in the world. pcTattletale’s website says the app “runs invisibly in the background on their workstations and can not be detected.”

But the bug means that anyone on the internet who understands how the security flaw works can download the screenshots captured by the spyware directly from pcTattletale’s servers. 

…It’s not known who planted the app or how the app was planted — for example, if hotel employees were tricked into installing it, or if the hotel owner intended the spyware to be used to monitor employee behavior.

«

unique link to this extract

Media companies are making a huge mistake with AI • The Atlantic

Jessica Lessin:

»

For as long as I have reported on internet companies, I have watched news leaders try to bend their businesses to the will of Apple, Google, Meta, and more. Chasing tech’s distribution and cash, news firms strike deals to try to ride out the next digital wave. They make concessions to platforms that attempt to take all of the audience (and trust) that great journalism attracts, without ever having to do the complicated and expensive work of the journalism itself. And it never, ever works as planned.

Publishers like News Corp did it with Apple and the iPad, investing huge sums in flashy content that didn’t make them any money but helped Apple sell more hardware. They took payouts from Google to offer their journalism for free through search, only to find that it eroded their subscription businesses. They lined up to produce original video shows for Facebook and to reformat their articles to work well in its new app. Then the social-media company canceled the shows and the app. Many news organizations went out of business.

The Wall Street Journal recently laid off staffers who were part of a Google-funded program to get journalists to post to YouTube channels when the funding for the program dried up. And still, just as the news business is entering a death spiral, these publishers are making all the same mistakes, and more, with AI.

Publishers are deep in negotiations with tech firms such as OpenAI to sell their journalism as training for the companies’ models. It turns out that accurate, well-written news is one of the most valuable sources for these models, which have been hoovering up humans’ intellectual output without permission. These AI platforms need timely news and facts to get consumers to trust them. And now, facing the threat of lawsuits, they are pursuing business deals to absolve them of the theft. These deals amount to settling without litigation. The publishers willing to roll over this way aren’t just failing to defend their own intellectual property—they are also trading their own hard-earned credibility for a little cash from the companies that are simultaneously undervaluing them and building products quite clearly intended to replace them.

«

Lessin is a founder of The Information, which charges as much – more? – than most national papers for its paywall. And is going from strength to strength, unlike a lot of media. But not everyone can do that.
unique link to this extract

I went undercover as a secret OnlyFans chatter. It wasn’t pretty • WIRED

Brendan Koerner:

»

To seal the deal, I needed to pass an elaborate written test. Daniel sent me a biographical sketch for a fictional “adult influencer from Tokyo” named Miko; she was a fan of karate, green tea, and the tongue emoji. My assignment was to write four extended back-and-forth dialogs between Miko and a hypothetical subscriber—two had to involve X-rated material, while the other two were meant to be clean. “Each bot’s reply should contain a call to action, a question, a compliment, or an inspiration to do something,” the instructions dictated, though I was forbidden from using question marks in more than 20% of Miko’s responses.

I found it quite easy at first to write the sort of run-of-the-mill smut the Serbs expected. (I’ll spare you the gory details, except to say I cribbed some color from Kathryn Bigelow’s 1995 sci-fi film Strange Days.) For the less explicit chats, I imagined Miko offering to cook the subscriber a pasta dinner and feigning appreciation for his TV recommendations. I did make one glaring error that could have led to an entire chat being voided as unusable: Due to my hasty misreading of Miko’s bio, I characterized her as a fan of spicy ramen when she actually prefers her food mild. “I have to ask you to pay attention to these little facts,” Daniel wrote in his assessment. “In this case, these lines mentioning the food could have been rejected, and that could have led to the dialog’s rejection.”

But despite that mistake and a few other hiccups—my punctuation seemed unnatural because it was too accurate—Daniel offered me the job. I was to be paid 7 cents per line of dialog, with each dialog running for a minimum of 40 lines. For my first assignment, I had to compose 20 dialogs involving sex in public places—10 at the beach, five inside a car, and five in a forest or garden. There was a list of particular sex acts I had to include, as well as a stricture that I refrain from using emoji in more than 30% of lines. I had only 48 hours to complete the task.

By the time I wrapped up my fifth dialog, my brain was a puddle of goo.

«

It turns out there’s an enormous hinterland of this stuff – agents and hiring and recruitment and levels and schemes – but when you weigh it up, all nonsense. So much human effort, for what?
unique link to this extract

Millennials are waving away privacy concerns and eyeing Chinese cars • Quartz

William Gavin:

»

The US Department of Commerce in February announced an investigation into Chinese-made smart cars, noting that EVs and autonomous vehicles collect massive amounts of data.

About 44% of all respondents [to a survey] said they would be very concerned about their privacy if Chinese cars were sold in the US, with another 34% saying they would be somewhat concerned, according to AutoPacific. But that won’t stop Americans under 40 from considering buying one of those vehicles, despite 73% saying they would be concerned about their privacy.

“Privacy concerns about Chinese-brand vehicles are likely to eventually subside given that most of the connected smartphones, smart watches, laptops, connected home devices we are comfortable using every day are in fact manufactured in China,” [chief analyst and president of research firm AutoPacific, Ed] Kim said.

Chinese cars — especially EVs — have a major leg up on Western competitors. Many Chinese automakers — thanks to a combination of cheap manufacturing and state subsidies — are able to sell their cars for dirt-cheap.

BYD, for example, sells the Seagull hatchback car for less than $10,000. China is also home to the cheapest electric car in the world, the roughly $1,000 Changli Freeman. Chinese cars also come with impressive technology and features that consumers demand, from better driver assistance tech to free fridges and portable rooms.

“Younger generations of shoppers are clearly aware of the enticing products Chinese automakers are cooking up overseas,” Robby DeGraff, AutoPacific’s manager of product and consumer insights, said in a statement. “It’s only a matter of ‘when’ they’ll be able to get their hands on them.”

«

unique link to this extract

Who actually uses Instagram’s Threads app? Taiwanese protestors • Rest of World

Viola Zhou:

»

As thousands of people gathered outside Taiwan’s legislature on Tuesday to protest against a bill that would give more power to China-friendly parties, Yuan, who was volunteering at a nearby church, noticed that the large crowd was running short on supplies. 

He fired off posts on the Threads app listing items that protesters needed, such as snacks, bottled water, and plastic bags. Supplies arrived within minutes. 

“My Threads page was like a wishing well,” Yuan, who requested to be identified with part of his first name for privacy reasons, told Rest of World. “We got everything we asked for.” 

A 32-year-old bar owner in Taipei, Yuan has been lurking on Threads since Meta launched the Instagram-linked alternative to X last year. He posted on the app for the first time last weekend to help organize a protest against the island’s opposition lawmakers. His posts about the protests have been “liked” thousands of times. 

Threads, which had 150 million monthly active users globally by April, is doing exceptionally well in Taiwan, where it’s commonly loosely transliterated as cui — because the “th” sound doesn’t appear in Mandarin. It works like X, allowing users to post 500-character-long text posts as well as audio, photos and short videos. Despite its small population of 23 million, Taiwan had 1.88 million active users on Threads from May 5 to 11, behind only the U.S., Japan, and Brazil, according to app-tracking site Data.ai. 

«

unique link to this extract

Apple’s WWDC may include AI-generated emoji and an OpenAI partnership • The Verge

Wes Davis:

»

Apple will finally tell its own AI story at WWDC 2024, but it may not mean the sorts of showy features demoed by the likes of Google, Microsoft, or OpenAI. Instead, the event may see Apple rolling out basic AI features like transcribing voice memos or auto-generated emoji — and announcing a rumored partnership with OpenAI, according to Mark Gurman’s Power On newsletter for Bloomberg today.

…The Voice Memo app could also get a big boost in AI-generated transcripts, Gurman writes. Selfishly, that will be key for referring to interview recordings, but it could also be handy for, say, students recording their lessons for later reference. Apple devices have similar features already, like auto-generated voicemail transcripts and system-wide captions for videos, audio, and conversations.

The company also reportedly plans to announce AI-powered improvements to on-device Spotlight search, internet searches with Safari, as well as writing suggestions for emails and texts. And the company may also use AI to retouch photos and generate emoji on the fly, based on what you’re texting — a type of feature that seems to consistently lead to trouble for these companies. (See Meta’s gun-toting Waluigi AI stickers or Google’s inappropriately racially diverse nazi pictures.)

Apple could showcase a better, more natural-sounding voice for Siri, based on Apple’s own large language models, as well as better Siri functionality on the Apple Watch.

«

unique link to this extract

California can make climate polluters pay for the mess they have made of Earth • Los Angeles Times

»

As climate change exacts a mounting toll on California, who should pay for the damage from rising sea levels and increasingly ferocious wildfires, floods and heat waves?

Fossil fuel companies would like taxpayers to keep footing the bill while they reap the profits from the burning of coal, oil and methane gas. That’s not right. Companies whose products are responsible for the vast majority of the greenhouse gas emissions should be held liable for the costs.

That’s the concept behind the Polluters Pay Climate Cost Recovery Act, a bill in the California Legislature that would create a Superfund-style program to collect money from major fossil fuel companies such as Chevron and ExxonMobil to help the state pay for the environmental damage caused by their products.

Other states, including New York, Massachusetts and Maryland, are considering similar “Climate Superfund” bills. They’re modeled on the 1980 federal Superfund law that established an industry-funded trust fund to pay for cleaning abandoned hazardous waste sites and holds current and past operators and other responsible parties liable for the costs. California would not be the first to act; Vermont lawmakers earlier this month sent a Climate Superfund bill to their governor’s desk.

…The fund would come from the estimated 41 fossil fuel extraction and refining companies that meet the bill’s threshold of being responsible for more than 1 billion metric tons of emissions from 2000 to 2020. Each would pay a share of the climate costs to the state based on a study to be conducted by the California Environmental Protection Agency. The total cost recovery could amount to tens or even hundreds of billions of dollars, according to legislative analysts, and could be paid in installments over 20 years.

«

You would feel fairly sure that the monies raised wouldn’t really go into climate amelioration – though taxes are rarely ringfenced to their purposed. This tax is overdue, to be honest. (This was an Editorial opinion in the paper, hence no byline.)
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

New documents from OpenAI claim to show that it did not copy Scarlett Johansson’s voice. But is that enough to head off her lawyers? CC-licensed photo by Chairman of the Joint Chiefs of Staff on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

It’s Friday, so there’s another post due at the Social Warming Substack at about 0845 UK time. On electioneering!

A selection of 9 links for you. I can hear you now. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

Google promised better search — now it’s telling us to put glue on pizza • The Verge

Kylie Robison:

»

Imagine this: you’ve carved out an evening to unwind and decide to make a homemade pizza. You assemble your pie, throw it in the oven, and are excited to start eating. But once you get ready to take a bite of your oily creation, you run into a problem — the cheese falls right off. Frustrated, you turn to Google for a solution.

“Add some glue,” Google answers. “Mix about 1/8 cup of Elmer’s glue in with the sauce. Non-toxic glue will work.”

So, yeah, don’t do that. As of writing this, though, that’s what Google’s new AI Overviews feature will tell you to do. The feature, while not triggered for every query, scans the web and drums up an AI-generated response. The answer received for the pizza glue query appears to be based on a comment from a user named “fucksmith” in a more than decade-old Reddit thread, and they’re clearly joking.

This is just one of many mistakes cropping up in the new feature that Google rolled out broadly this month. It also claims that former US President James Madison graduated from the University of Wisconsin not once but 21 times, that a dog has played in the NBA, NFL, and NHL, and that Batman is a cop. [Also that Barack Obama was a Muslim US president – Overspill Ed.]

Look, Google didn’t promise this would be perfect, and it even slaps a “Generative AI is experimental” label at the bottom of the AI answers. But it’s clear these tools aren’t ready to accurately provide information at scale.

«

Look, these tools will never be ready to “accurately provide information at scale”. On Thursday night I tried the “African countries beginning with K” query, which it was getting wrong in November 2023, and it still gets it wrong. Somehow, people at Google thinks the internet is full of correct information. That suggests they’ve never actually met the internet.

This product is not fit for purpose, and never can be as long as the internet contains errors. Which means forever.
unique link to this extract

Sky voice actor says nobody ever compared her to ScarJo before OpenAI drama • Ars Technica

Ashley Belanger:

»

OpenAI is sticking to its story that it never intended to copy Scarlett Johansson’s voice when seeking an actor for ChatGPT’s “Sky” voice mode.

The company provided The Washington Post with documents and recordings clearly meant to support OpenAI CEO Sam Altman’s defense against Johansson’s claims that Sky was made to sound “eerily similar” to her critically acclaimed voice acting performance in the sci-fi film Her.

Johansson has alleged that OpenAI hired a soundalike to steal her likeness and confirmed that she declined to provide the Sky voice. Experts have said that Johansson has a strong case should she decide to sue OpenAI for violating her right to publicity, which gives the actress exclusive rights to the commercial use of her likeness.

In OpenAI’s defense, The Post reported that the company’s voice casting call flier did not seek a “clone of actress Scarlett Johansson,” and initial voice test recordings of the unnamed actress hired to voice Sky showed that her “natural voice sounds identical to the AI-generated Sky voice.” Because of this, OpenAI has argued that “Sky’s voice is not an imitation of Scarlett Johansson.”

What’s more, an agent for the unnamed Sky actress who was cast—both granted anonymity to protect her client’s safety—confirmed to The Post that her client said she was never directed to imitate either Johansson or her character in Her. She simply used her own voice and got the gig.

The agent also provided a statement from her client that claimed that she had never been compared to Johansson before the backlash started.

This all “feels personal,” the voice actress said, “being that it’s just my natural voice and I’ve never been compared to her by the people who do know me closely.”

However, OpenAI apparently reached out to Johansson after casting the Sky voice actress. During outreach last September and again this month, OpenAI seemed to want to substitute the Sky voice actress’s voice with Johansson’s voice—which is ironically what happened when Johansson got cast to replace the original actress hired to voice her character in Her.

«

Either there’s going to be a big payout to Johansson, or this is all going to court. The giveaway is the initial approach to Johansson, and Altman’s tweeting of “her” during the demo. There’s absolutely no mistaking the intention.
unique link to this extract

Indian voters are being bombarded with millions of deepfakes. Political candidates approve • WIRED

Nilesh Christopher and Varsha Bansal:

»

On a stifling April afternoon in Ajmer, in the Indian state of Rajasthan, local politician Shakti Singh Rathore sat down in front of a greenscreen to shoot a short video. He looked nervous. It was his first time being cloned.

Wearing a crisp white shirt and a ceremonial saffron scarf bearing a lotus flower—the logo of the BJP, the country’s ruling party—Rathore pressed his palms together and greeted his audience in Hindi. “Namashkar,” he began. “To all my brothers—”

Before he could continue, the director of the shoot walked into the frame. Divyendra Singh Jadoun, a 31-year-old with a bald head and a thick black beard, told Rathore he was moving around too much on camera. Jadoun was trying to capture enough audio and video data to build an AI deepfake of Rathore that would convince 300,000 potential voters around Ajmer that they’d had a personalized conversation with him—but excess movement would break the algorithm. Jadoun told his subject to look straight into the camera and move only his lips. “Start again,” he said.

Right now, the world’s largest democracy is going to the polls. Close to a billion Indians are eligible to vote as part of the country’s general election, and deepfakes could play a decisive, and potentially divisive, role. India’s political parties have exploited AI to warp reality through cheap audio fakes, propaganda images, and AI parodies. But while the global discourse on deepfakes often focuses on misinformation, disinformation, and other societal harms, many Indian politicians are using the technology for a different purpose: voter outreach.

Across the ideological spectrum, they’re relying on AI to help them navigate the nation’s 22 official languages and thousands of regional dialects, and to deliver personalized messages in farther-flung communities. While the US recently made it illegal to use AI-generated voices for unsolicited calls, in India sanctioned deepfakes have become a $60m business opportunity.

«

Strange to think of deepfakes being used positively, but that really is the case here.
unique link to this extract

Exclusive: Google parent Alphabet weighs offer for HubSpot • Reuters

Anirban Sen and Milana Vinn:

»

Google parent Alphabet has been talking to its advisers about the possibility of making an offer for HubSpot, an online marketing software company with a market value of $35bn, people familiar with the matter said.

If Alphabet moves ahead with a bid, it would be a rare example of a major technology company attempting a mega deal amid heightened regulatory scrutiny of the sector under U.S. President Joe Biden’s administration.

The potential acquisition would be Alphabet’s largest ever and allow it to put some of its cash pile, which reached $110.9bn at the end of December, to work.

Alphabet has met with Morgan Stanley (MS.N), opens new tab investment bankers in recent days about a potential offer for HubSpot, the sources said. It has been discussing how much it should offer and whether antitrust regulators would clear such a tie-up, the sources added.

Alphabet has not yet submitted an offer to HubSpot and there is no certainty it will do so, the sources said, requesting anonymity to discuss confidential deliberations.

“As standard practice, HubSpot does not comment on rumors or speculation. We continue to focus on building a great business and serving our customers,” a HubSpot spokesperson said.

«

HubSpot is the company where Dan Lyons, former journalist, worked and found life so ridiculous that he wrote the book Disrupted – which is hilarious. Though I think it’s the HubSpot shareholders who would be laughing all the way to the bank if this comes off.

Why does Google want it? Apparently, it “would expand Google’s offerings in the booming market for customer relationship management (CRM) software, enabling it to tap a wider base of enterprise customers who spend on marketing and advertising.” Can’t see this passing antitrust examination, to be honest. Or shouldn’t.
unique link to this extract

Samsung requires independent repair shops to share customer data, snitch on people who use aftermarket parts, leaked contract shows • 404 Media

Jason Koebler:

»

In exchange for selling them repair parts, Samsung requires independent repair shops to give Samsung the name, contact information, phone identifier, and customer complaint details of everyone who gets their phone repaired at these shops, according to a contract obtained by 404 Media. Stunningly, it also requires these nominally independent shops to “immediately disassemble” any phones that customers have brought them that have been previously repaired with aftermarket or third-party parts and to “immediately notify” Samsung that the customer has used third-party parts.

…The contract also requires the “daily” uploading of details of each and every repair that an independent company does into a Samsung database called G-SPN “at the time of each repair,” which includes the customer’s address, email address, phone number, details about what is wrong with their phone, their phone’s warranty status, details of the customer’s complaint, and the device’s IMEI number, which is a unique device identifier. 404 Media has verified the authenticity of the original contract and has recreated the version embedded at the bottom of this article to protect the source. No provisions have been changed.

The use of aftermarket parts in repair is relatively common. This provision requires independent repair shops to destroy the devices of their own customers, and then to snitch on them to Samsung.

«

I don’t think even Apple would go this far.
unique link to this extract

Three bullet points: internet time ain’t what it used to be • The Future, Now and Then

Dave Karpf:

»

I think one reason why people in my age bracket have such strong, implicit faith in Moore’s Law is that it was part of our shared reality for such a long time. Consumer tech really was getting significantly better and significantly cheaper, at a pace that you could not help but notice.

I saved up all summer in 1998 to buy a nice stereo. (It played tapes AND cds!) Four years later, my friend Becca was showing off the clickwheel on her new iPod.

In 2021, the keyboard on my laptop started having trouble. The “e” key stopped working. I checked, and found the computer wasn’t under warranty anymore. Turns out I had bought it way back in 2012. It still worked fine, except for the damn “e” key.

When I started college in 1997, a nine year old computer (from 1988!) would, for all practical purposes, not be a computer at all.

Silicon Valley’s aura of futurity was honed through this frenzied cycle of consumer product upgrades. The leveling-off in consecutive iterations of Apple’s product lines today was nowhere to be seen. There was forever a next generation of consumer products coming, and that next generation was demonstrably better and cheaper than the one that preceded it.

It felt just a bit magical. No other part of the physical world was transforming at such a constant, reliable pace. But this is no longer the case, and it hasn’t been for quite awhile.

And in the meantime, the mythos surrounding Moore’s Law keeps being propped up, Weekend at Bernie’s-style.

Sure, the Rabbit R1 might be utter trash. Yes, the Humane AI pin is so bad it’s unreviewable. But why be so dour about these early models? Just focus on how much better the next one will be. Surely it’s just around the corner. (Because Moore’s Law!)

«

Karpf’s point is that Moore’s Law only applied narrowly: to chip manufacture. A generation that grew up with it thinks that will apply too on products like LLMs. But it very much doesn’t and won’t.
unique link to this extract

The price of computer storage has fallen exponentially since the 1950s • Our World in Data

Edouard Mathieu:

»

This chart [in the article, showing log price per terabyte over time] shows the dramatic fall in the price of computer storage between 1956 and 2023. It relies on the data carefully collected by the computer scientist John C. McCallum.

In the past 70 years, the price for a unit of storage has fallen by almost ten orders of magnitude. The data is plotted on a logarithmic scale on the vertical axis. The line follows an almost straight path, indicating an exponential reduction in price.

A 256-gigabyte storage capacity — commonly found in standard laptops sold today — would have cost around $20bn in the 1950s. (That’s in today’s prices.)

And cost has not been the only improvement: modern solid-state drives offer much faster and more reliable data access than early magnetic and hard disk drives.

«

Intriguingly, “flash” memory is recorded from 2003 but there’s a separate category of “solid state” from 2013. (That distinction is in the original data too.) But flash memory is solid state memory. Anyhow, the prices shown are at retail (once storage was available from retail, in about 1975). There’s also another potential axis: speed. A 3D graphic showing that would be equally amazing.
unique link to this extract

Political consultant behind fake Biden AI robocall faces charges in New Hampshire • CNN Politics

Rashard Rose and Marshall Cohen:

»

New Hampshire prosecutors filed 26 criminal charges against the political consultant behind a robocall that used artificial intelligence to impersonate President Joe Biden and urged voters not to participate in the state’s primary this year.

The Federal Communications Commission also imposed a $6m fine against the consultant, Steve Kramer, because the robocalls used call-spoofing technology that violated federal caller-ID laws.

Kramer, 54, was named in several indictments in different New Hampshire counties, according to court documents obtained by CNN. He faces 13 charges of felony voter intimidation or suppression. He also faces 13 counts of impersonating a candidate, which is a misdemeanor, according to the court documents.

The indictments, filed by the New Hampshire attorney general, allege that Kramer “sent or caused to be sent a pre-recorded phone message that disguised the source of the call, or was deceptive in using an artificially created voice of a candidate, or provided misleading information, in attempting to prevent or deter” voters from participating in the New Hampshire primary.

«

That’s quite a deterrent for anyone wanting to do some voter suppression.
unique link to this extract

Spotify is going to break every Car Thing gadget it ever sold • The Verge

Chris Welch:

»

Spotify’s brief attempt at being a hardware company wasn’t all that successful: the company stopped producing its Car Thing dashboard accessory less than a year after it went on sale to the public. And now, two years later, the device is about to be rendered completely inoperable. Customers who bought the Car Thing are receiving emails warning that it will stop working altogether as of December 9th.

Unfortunately for those owners, Spotify isn’t offering any kind of subscription credit or automatic refund for the device — nor is the company open-sourcing it. Rather, it’s just canning the project and telling people to (responsibly) dispose of Car Thing.

“We’re discontinuing Car Thing as part of our ongoing efforts to streamline our product offerings,” Spotify wrote in an FAQ on its website. “We understand it may be disappointing, but this decision allows us to focus on developing new features and enhancements that will ultimately provide a better experience to all Spotify users.”

…The Car Thing hardware was quite nice considering it was Spotify’s first go, but the product was more of a remote control for Spotify on your mobile phone than any kind of standalone player.

«

Wave goodbye to the $90 spent on it. But then again, it was only produced for five months. Originally described as:

»

Car Thing [is] a Spotify-only, voice-controlled device for the car, is launching today in limited quantities to invited users. It’s a dedicated, Bluetooth-connected device for controlling Spotify without the need for a phone screen, which seems to be meant for people who drive older cars without built-in infotainment systems or phone connections.

«

Conclusion: there aren’t that many Spotify users driving cars without built-in infotainment or phone connections. Failure of market research, basically.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

Most cars in Ethiopia are very old – but a government scheme is pushing electric vehicles, as electricity is cheap and oil expensive. CC-licensed photo by Rachel Strohm on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 10 links for you. Use them wisely. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

Microsoft Copilot+ Recall feature is a ‘privacy nightmare’ • BBC News

Imran Rahman-Jones:

»

The UK data watchdog says it is “making enquiries with Microsoft” over a new feature that can take screenshots of your laptop every few seconds.

Microsoft says Recall, which will store encrypted snapshots locally on your computer, is exclusive to its forthcoming Copilot+ PCs. But the Information Commissioner’s Office (ICO) says it is contacting Microsoft for more information on the safety of the product, which privacy campaigners have called a potential “privacy nightmare”.

Microsoft says Recall is an “optional experience” and it is committed to privacy and security. According to its website, external, users “can limit which snapshots Recall collects”.

“Recall data is only stored locally and not accessed by Microsoft or anyone who does not have device access,” the firm said in a statement. And it said a would-be hacker would need to gain physical access to your device, unlock it and sign in before they could access saved screenshots.

But an ICO spokesperson said firms must “rigorously assess and mitigate risks to peoples’ rights and freedoms” before bringing any new products to market. “We are making enquiries with Microsoft to understand the safeguards in place to protect user privacy,” they said.

Recall has the ability to search through all users’ past activity including files, photos, emails and browsing history. Many devices can already do this – but Recall also takes screenshots every few seconds and searches these too.

“This could be a privacy nightmare,” said Dr Kris Shrishak, an adviser on AI and privacy. “The mere fact that screenshots will be taken during use of the device could have a chilling effect on people.”

«

Recall does feel like a half-done solution to a tricky problem. At the very least you’d want the screenshots to be protected behind a password that isn’t the same as the login password, and ideally encrypted in the same way.
unique link to this extract

Humane is looking for a buyer after the AI Pin’s underwhelming debut • The Verge

Chris Welch:

»

Humane, the startup behind the poorly-reviewed AI Pin wearable computer, is already hunting for a potential buyer for its business. That’s according to a report from Bloomberg, which says the company — led by former longtime Apple employees Imran Chaudhri and Bethany Bongiorno — is “seeking a price of between $750m and $1bn.”

That might be a tough sell after the $699 AI Pin’s debut: the device has been widely panned for its slow responses and a user experience that falls well short of the always-on, wearable AI assistant concept that its founders promised in the run-up to the device’s release. The product was pitched at least partially as a way for people to be more present and reduce their ever-growing dependence on smartphones.

Humane developed its own operating system called CosmOS that runs on the AI Pin. It hooks into a network of AI models to fetch answers for voice queries and to analyze what the built-in camera is pointed at. For some interactions, the device beams out a laser “display” that is projected onto the wearer’s inner palm. A monthly subscription is required to keep the device active.

The Bloomberg report notes that Humane has raised $230m from investors including OpenAI CEO Sam Altman, who is rumored to be developing an unrelated product (in collaboration with legendary Apple designer Jony Ive) that could better showcase AI’s promise.

«

Overpriced, even in its death throes. What do we think: $50m and an acquihire in a couple of months?
unique link to this extract

Google Search’s new AI overviews will soon have ads • WIRED

Paresh Dave:

»

Last week Google introduced a radical shake-up of search that presents users with AI-generated answers to their queries. Now the company says it will soon start including ads inside those AI Overviews, as the automatic answers are called.

Google on Tuesday announced plans to test search and shopping ads in the AI summaries, a move that could extend its dominance in search advertising into a new era. Although Google rapidly rolled out AI Overviews to all US English users last week after announcing the feature at its I/O developer conference, it’s unclear how widely or quickly ads will start appearing.

Screenshots released by Google show how a user asking how to get wrinkles out of clothes might get an AI-generated summary of tips sourced from the web, with a carousel of ads underneath for sprays that purport to help crisp up a wardrobe.

Google’s AI Overviews are meant to keep users from shifting to alternatives such as ChatGPT or the startup Perplexity, which use AI-generated text to answer many questions traditionally thrown at Google. How and when Google would integrate ads into AI Overviews has been a significant question over the company’s ChatGPT catch-up strategy. Search ads are the company’s largest revenue generator, and even subtle changes in ad placements or design can spur big swings in Google’s revenue.

Google shared few details about its new Overview ad format in its announcement Tuesday. Ads “will have the opportunity to appear within the AI Overview in a section clearly labeled as ‘sponsored’ when they’re relevant to both the query and the information in the AI Overview,” Vidhya Srinivasan, Google’s vice president and general Manager for ads, wrote in a blog post.

AI Overview will draw on ads from advertisers’ existing campaigns, meaning they can neither completely opt out of the experiment nor have to adapt the settings and designs of their ads to appear in the feature.

«

Srinivasan was described by Ed Zitron as “The Man Who Killed Google Search“, based on documents released at the DoJ trial which showed how more and more ads were packed into search results to hit financial targets. Seems like this is the continuation: strangle the web in favour of adverts.
unique link to this extract

Connected fitness is adrift post-pandemic • TechCrunch

Brian Heater, amid the news that Peloton is looking to refinance its debt while taking on a $1bn five-year loan:

»

The pandemic has certainly had long-term impacts on the economy. For instance, while work from home has obviously declined from its COVID heights, a report earlier this year notes that it’s still in the area of three to four times more common than it was in 2019. Connected fitness’ big bet was that while some regression was inevitable, the cultural shift was going to be permanent.

Ultimately, however, many were eager for a “return to normality,” and arrival of vaccinations, coupled with lowered rates of infection, emboldened many to get back to the gym. Unlike commuting into an artificially lit cubical farm five times a week, plenty of people genuinely enjoy the experience of working out in person.

The struggle isn’t universal, however. Hydrow, which raised $55m in 2022, purchased a majority stake in AI-based strength training firm Speede Fitness earlier this month. The firm has done a good job capitalizing on interest around rowing machines, even as Peloton’s answer to the category was entirely overshadowed by its very public struggles.

Despite some major regressions and broader economic headwinds, there’s always money to be raised if you’ve got a compelling enough product. Ultimately, however, those rounds should be consistently lower than they were in the home fitness salad days. For a recent example, Kabata, the maker of the “world’s first AI-powered dumbbells,” announced a $5m seed round on Tuesday. That’s follows a $2m seed round raised in May 2022.

«

AI-powered dumbbells. The website makes lots of promises. They’re not on sale yet (autumn!), require a subscription, iOS only, require Bluetooth and Wi-Fi, US only.

How much? $800 outright. Suggestion to American fitness readers: buy normal dumbbells, get a personal trainer.
unique link to this extract

‘Climate change is on the ballot’: Prime Minister calls surprise summer UK election • BusinessGreen News

Michael Holder and James Murray:

»

Government Ministers have repeatedly argued Labour’s plan to deliver a clean power system by 2030 is not feasible and risks driving up energy bills, while also highlighting their record of delivering the deepest decarbonisation in the G20 and engineering a huge increase in renewables capacity and green investment.

But Labour has hit back by accusing the government of overseeing policy uncertainty and insufficient infrastructure underinvestment that has hampered the clean energy transition, driven up energy bills, and led to a sewage pollution crisis.   

Green groups today urged both main parties to prioritise ambitious climate action in the coming election.

“The election has been called on the day that scientists announced that “never ending” rain in the UK in autumn and winter has been made 10 times more likely by the climate crisis,” said Ed Matthew, campaigns director at climate change think tank E3G. “Climate change is already causing devastation to crops and homes in the UK and our dependence on fossil fuels has been at the heart of the cost of living crisis. Climate change is on the ballot like never before and voters will be looking for bold manifesto pledges to show parties are committed to ambitious action to rapidly build a clean, green energy system. Any party failing to take action on climate change is condemning itself to electoral oblivion.”

Leo Murray, co-director of climate charity Possible, urged political leaders to avoid the temptation to stoke divisions on environmental issues.

“We can’t let climate change become a culture war in this general election campaign,” he said. “During the 2019 general election, it was an issue on the doorstep for the first time. The first ever televised climate debate was broadcast into millions of living rooms and leading politicians pledged billions for climate action. But going into this election, the political consensus around climate is in peril.  Government ministers peddle conspiracy theories. Our prime minister has watered down net zero targets. Politicians are trying to turn climate into a dividing line, rather than a way to bring people together.”

«

Related: heavy rainfall expected to lead to £1bn fall in arable farm revenues – which will probably push up inflation in the autumn. By which time it’ll be Someone Else’s Problem, as far as the Tories are concerned.
unique link to this extract

Ethiopia shows us just how fast the transition to electric mobility can happen in Africa • CleanTechnica

Remeredzai Joseph Kuhudzai:

»

Like a lot of countries on the African continent, Ethiopia has an exceptionally low motorisation rate. Ethiopia has a population of 126 million people, but the total number of vehicles registered in Ethiopia is around 1.2 million. Most of these vehicles are over 20 years old.

According to reports and announcements from the Ethiopian government, Ethiopia had a plan to catalyse adoption of electric vehicles in Ethiopia with a 10 year target to see 148,000 electric cars and close to 50,000 electric buses on Ethiopia’s roads by 2030. However, Ethiopia’s Ministry of Transport and Logistics recently said that this target of over 100,000 EVs has already been met in just the first two years of this plan.

Due to this incredible progress, the target has since been bumped up to close to 500,000 in the 10-year period. So, in just two years, locally assembled EVs and imported EVs have added almost 10% of Ethiopia’s current total ICE vehicle registrations. If all the vehicles in the current fleet stay on the road for the next 8 years (highly unlikely), the total fleet will be 1.7 million. If the target is met, it would mean the penetration of electric vehicles in Ethiopia’s total fleet will be close to 30% at that time.

Of course, the actual number will be more than 30%, as a lot of the vehicles in the current ICE fleet will be retired by then. Also, given the extremely low motorisation in Ethiopia, vehicle sales should grow at a much faster rate going forward, and probably the penetration of EVs in the country’s total fleet will hit close to 50% by then.

«

And why the push for EVs?

»

Ethiopia spends over US$5bn annually on petrol and diesel imports, precious foreign currency it does not have. Ethiopia has also recently started generating electricity from the first units at the 5-gigawatt Grand Ethiopian Renaissance Dam, adding to its other hydro and renewable energy resources. Electricity is ridiculously cheap in Ethiopia at under 1¢ USD per kWh in many cases.

«

unique link to this extract

Microsoft’s AI obsession is jeopardizing its climate ambitions • The Verge

Justine Calma:

»

Back in 2020, Microsoft set a target of becoming carbon negative by the end of the decade. To translate the jargon, it pledged to slash greenhouse gas emissions by more than half and then capture a greater amount of carbon dioxide emissions than it would produce. It was an audacious commitment to make at the time, considering carbon capture technologies were barely coming into existence. The company would also have to spur the deployment of way more renewable energy onto power grids where it operates.

Now, it looks like the company’s recent obsession with AI is making that much harder to achieve. Microsoft has invested more than $13bn in OpenAI to date, and it’s “turning everyone into a prompt engineer” for generative AI with new features in Copilot for Microsoft 365. (The Verge’s Tom Warren just launched a newsletter called Notepad to keep you up to date on all things Microsoft and AI.)

“In 2020, we unveiled what we called our carbon moonshot. That was before the explosion in artificial intelligence,” Microsoft president Brad Smith said in an interview with Bloomberg. “So in many ways the moon is five times as far away as it was in 2020, if you just think of our own forecast for the expansion of AI and its electrical needs.”

«

But on the plus side, you have autocomplete for that code to make a box of advertising pop up in the middle of the screen when someone’s reading a story, so net-net about even?
unique link to this extract

Risk assessment of a highly pathogenic H5N1 influenza virus from mink • Nature Communications

Katherine Restori et al:

»

Outbreaks of highly pathogenic H5N1 clade 2.3.4.4b viruses in farmed mink and seals combined with isolated human infections suggest these viruses pose a pandemic threat. To assess this threat, using the ferret model, we show an H5N1 isolate derived from mink transmits by direct contact to 75% of exposed ferrets and, in airborne transmission studies, the virus transmits to 37.5% of contacts. Sequence analyses show no mutations were associated with transmission.

The H5N1 virus also has a low infectious dose and remains virulent at low doses. This isolate carries the adaptive mutation, PB2 T271A, and reversing this mutation reduces mortality and airborne transmission. This is the first report of a H5N1 clade 2.3.4.4b virus exhibiting direct contact and airborne transmissibility in ferrets. These data indicate heightened pandemic potential of the panzootic H5N1 viruses and emphasize the need for continued efforts to control outbreaks and monitor viral evolution.

«

Watching brief, nothing more. Cross-species transmission without laboratory manipulation, you say? Unimaginable.
unique link to this extract

Eventbrite promoted illegal opioid sales to people searching for addiction recovery help • WIRED

Matt Burgess and Dhruv Mehrotra:

»

This June, approximately 150 motorcycles will thunder down Route 9W in Saugerties, New York, for Ryan’s Ride for Recovery. Organized by Vince Kelder and his family, the barbecue and raffle will raise money to support their sober-living facility and honor their son who tragically died from a heroin overdose in 2015 after a years-long drug addiction.

The Kelders established Raising Your Awareness about Narcotics (RYAN) to help others struggling with substance-use disorder. For years, the organization has relied on Eventbrite, an event management and ticketing website, to arrange its events. This year, however, alongside listings for Ryan’s Ride and other addiction recovery events, Eventbrite surfaced listings peddling illegal sales of prescription drugs like Xanax, Valium, and oxycodone.

“It’s criminal,” Vince Kelder says. “They’re preying on people trying to get their lives back together.”

Eventbrite prohibits listings dedicated to selling illegal substances on its platform. It’s one of the 16 categories of content the company’s policies restrict its users from posting. But a WIRED investigation found more than 7,400 events published on the platform that appeared to violate one or more of these terms.

Among these listings were pages claiming to sell fentanyl powder “without a prescription,” accounts pushing the sale of Social Security numbers, and pages offering a “wild night with independent escorts” in India. Some linked to sites offering such wares as Gmail accounts, Google reviews (positive and negative), and TikTok and Instagram likes and followers, among other services.

«

Not only that, but the algorithm pushed those illicit listings to people trying to recover. Shall we guess? Could it be that Eventbrite doesn’t monitor the listings at all, just takes the money from them? What an incredible thought. Would it surprise you to hear that “Eventbrite appears to have removed most, if not all, of the illicit listings that WIRED identified after we alerted the company to the issue.” [Italics added.]
unique link to this extract

Nova explosion visible to the naked eye expected any day now • Ars Technica

Jacek Krywko:

»

When you look at the northern sky, you can follow the arm of the Big Dipper as it arcs around toward the bright star called Arcturus. Roughly in the middle of that arc, you’ll find the Northern Crown constellation, which looks a bit like a smiley face. Sometime between now and September, if you look to the left-hand side of the Northern Crown, what will look like a new star will shine for five days or so.

This star system is called T. Coronae Borealis, also known as the Blaze Star, and most of the time, it is way too dim to be visible to the naked eye. But once roughly every 80 years, a violent thermonuclear explosion makes it over 10,000 times brighter. The last time it happened was in 1946, so now it’s our turn to see it.

“The T. Coronae Borealis is a binary system. It is actually two stars,” said Gerard Van Belle, the director of science at Lowell Observatory in Flagstaff, Arizona. One of these stars is a white dwarf, an old star that has already been through its fusion-powered lifecycle. “It’s gone from being a main sequence star to being a giant star. And in the case of giant stars, what happens is their outer parts eventually get kind of pushed into outer space. What’s left behind is a leftover core of the star—that’s called a white dwarf,” Van Belle explained.

The white dwarf stage is normally a super peaceful retirement period for stars. The nuclear fusion reaction no longer takes place, which makes white dwarfs very dim. They are still pretty hot, though, and they’re super dense, with a mass comparable to our Sun squeezed into a volume resembling the Earth.

But the retirement of the white dwarf in T. Coronae Borealis is hardly peaceful, as it has a neighbor prone to littering. “Its companion star is in the red giant phase, where it is puffed up. Its outer parts are getting sloughed off and pushed into space. The material that is coming off the red giant is now falling onto the white dwarf,” Van Belle said.

«

Just as long as it’s not our star going nova.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

Great spotted eagles changed their migration routes to avoid Ukraine once the conflict began there in 2022. CC-licensed photo by Bernard DUPONT on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 10 links for you. Like Steve Miller? I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

How the new Microsoft Recall feature fundamentally undermines Windows security • DoublePulsar

Kevin Beaumont:

»

On Monday, Microsoft CEO Satya Nadella sat down with the media to introduce a new feature called Recall, as part of their Copilot+ PCs. It takes screenshots of what you’re doing on constantly, by design:

The idea is it allows you to rewind back in time at the click of a button to see what you were doing at, say, 11pm two months ago. It also classifies almost everything you’re doing, seeing and typing. This is instantly searchable.

Microsoft’s store page for the PCs points out “It will not hide information such as passwords or financial account numbers [..]” For example, if you log into online banking, your information around account numbers, balances, purchases etc will enter Recall’s database.

This fundamentally changes the relationship with you and your Microsoft Windows computer. It also introduces real risk to you, the customer. Let’s break down what is happening.

You may look at this and think ‘surely there’s some safety guardrails’, and there are — in the video above you’ll see Satya point out the processing and data storage is done locally on the device. In the FAQ they point out there’s some circumstances where data won’t be recorded, for example when password’s aren’t visible on screen.

That’s great. It’s also not nearly enough.

If you look at what has happened historically with infostealer malware — malicious software snuck onto PCs — it has pivoted to automatically steal browser passwords stored locally. In other words, if a malicious threat actor gains access to a system, they already steal important databases stored locally.

«

You can turn it off, but of course then.. you’ve turned it off. This is going to be a hell of a target for malware.
unique link to this extract

The tragic downfall of the internet’s art gallery • Slate

Nitish Pahwa:

»

On March 27, a large group of artists and creators from across the web noticed the frightening extent to which a once-beloved, highly influential community platform of theirs had, like so many others, fallen prey to the artificial intelligence juggernauts plundering the internet.

As VFX animator Romain Revert (Minions, The Lorax) pointed out on X, the bots had come for his old home base of DeviantArt. Its social accounts were promoting “top sellers” on the platform, with usernames like “Isaris-AI” and “Mikonotai,” who reportedly made tens of thousands of dollars through bulk sales of autogenerated, dead-eyed 3D avatars. The sales weren’t exactly legit—an online artist known as WyerframeZ looked at those users’ followers and found pages of profiles with repeated names, overlapping biographies and account-creation dates, and zero creations of their own, making it apparent that various bots were involved in these “purchases.”

It’s not unlikely, as WyerframeZ surmised, that someone constructed a low-effort bot network that could hold up a self-perpetuating money-embezzlement scheme: Generate a bunch of free images and accounts, have them buy and boost one another in perpetuity, inflate metrics so that the “art” gets boosted by DeviantArt and reaches real humans, then watch the money pile up from DeviantArt revenue-sharing programs. Rinse, repeat.

After Revert declared this bot-on-bot fest to be “the downfall of DeviantArt,” myriad other artists and longtime users of the platform chimed in to share in the outrage that these artificial accounts were monopolizing DeviantArt’s promotional and revenue apparatuses. Several mentioned that they’d abandoned their DeviantArt accounts—all appearing to prove his dramatic point.

«

unique link to this extract

Stalk my date: I know where you were last night • Bustle

Kate Lindsay:

»

Location-monitoring is all but expected in relationships nowadays, in part because it comes in so many forms — from Life360 to Snap Map. Whether they mean to or not, users are constantly signaling their availability or where they are in real time. Instagram beams out a bright green dot next to my username whenever I’m scrolling; Facebook Messenger allows me to toggle sharing my whereabouts. It’s so easy, people may not even know when and where they’re doing it. Just ask 31-year-old Carlotta*, whose ex seemed to have no idea he had turned location services on for Facebook Messenger while they were dating.

“It was definitely by accident,” she told me. “He would message me, ‘Just leaving my house for work. What are you doing tonight?’ I would see in his location services that he was somewhere completely different.” She used this information to confront him, and he admitted to sleeping with someone else.

For those with an iPhone, one location-sharing feature, Find My Friends, dominates. Apple launched its standalone app in 2011, and by 2015, it came automatically with new iPhones. Since it merged with Find My iPhone and Find My Mac in 2019, Find My has been a one-stop shop for all users’ surveillance needs. Even when I’m not actively checking in on my friends and family, my iPhone is: I’ll reliably get an alert when my boyfriend (or, really, his AirPods) have been near me for a while.

Just last week, my friend Selina Scharr, a 31-year-old from Brooklyn, dropped her dog off at the vet for a stressful surgery. After an emotional goodbye, she left, only to receive an alert from the AirTag she keeps on her dog’s collar. Basil, it warned her, had been “left behind.”

“Jeez, Apple, way to make me cry this morning,” she texted our group chat. (Basil is recovering happily at home.)

This technology has changed everything. I no longer have to ask my friend for their ETA or wonder if my parents are available for a phone call. I can just check in on their real-time location, a capability previously associated with intelligence agencies and covert criminals. What is, for all intents and purposes, stalking, has become a casual part of everyday familial, romantic, and platonic relationships — and turned many with formerly healthy boundaries into lurkers in the process.

«

unique link to this extract

Everyone in Hollywood is using AI, but “they are scared to admit it” • Hollywood Reporter

Winston Cho and Scott Roxborough:

»

For horror fans, Late Night With the Devil marked one of the year’s most anticipated releases. Embracing an analog film filter, the found-footage flick starring David Dastmalchian reaped praise for its top-notch production design by leaning into a ’70s-era grindhouse aesthetic reminiscent of Dawn of the Dead or Death Race 2000. Following a late-night talk show host airing a Halloween special in 1977, it had all the makings of a cult hit.

But the movie may be remembered more for the controversy surrounding its use of cutaway graphics created by generative artificial intelligence tools. One image of a dancing skeleton in particular incensed some theatergoers. Leading up to its theatrical debut in March, it faced the prospect of a boycott, though that never materialized.

The movie’s directors Cameron and Colin Cairnes defended the AI usage, explaining the art was touched up by human hands. In a statement, they said, “We experimented with AI for three still images which we edited further and ultimately appear as very brief interstitials in the film.”

Less than a month later, five images suspected to be generated by AI teasing postapocalyptic scenes in A24’s Civil War sparked similar outrage by a segment of fans. There were a few telltale signs that the graphics were AI-created in landmark accuracy and consistency blunders: The two Chicago Marina Towers buildings in one poster are on opposite sides of the river; in another, a shot of wreckage shows a car with three doors.

In response, a reader on A24’s Instagram post wrote that the backlash to Late Night was “more than enough to make transparently clear to everyone: WE DO NOT WANT THIS.” 

But in the entertainment industry, the Pandora’s box of AI has likely already been unleashed. Behind closed doors, most corners of production, from writers’ rooms to VFX departments, have embraced generative AI tools.

«

Those Civil War posters (in the linked article) do look weird, and AI-adjacent – especially the car with three doors on one side. They aren’t scenes from the (excellent) film itself.
unique link to this extract

Four more cats die of H5N1 bird flu in the US • BNO News

»

Four more cats have died of H5N1 bird flu in the United States, including two pets in South Dakota with no links to poultry or dairy cows, according to state and federal officials. At least 14 cats have recently died of bird flu.

Of the newly reported cases, two were domestic cats which died at a property in Campbell County in South Dakota, according to a state official and the U.S. Department of Agriculture (USDA).

Beth Thompson, the state’s veterinarian, said there was no livestock on the property where the pets died. “No other details regarding how the cats were infected are known at this time,” Thompson told BNO News.

Two other cases were recently reported in Michigan, one in Isabella County and the other in Ionia County. Both cases involved barn cats on commercial dairy farms where cows were also infected with H5N1.

At the property in Ionia County, two Virginia opossums were also infected with the bird flu virus.

“Cats are particularly susceptible to H5N1 2.3.4.4b viruses and the majority of sick cats have been reported at or near affected poultry facilities or dairies,” Shilo Weir, a spokesperson for USDA’s Animal and Plant Health Inspection Service, told BNO News.

…In the United States, at least 27 cats have now been infected with H5N1 bird flu, including the 14 cases reported in recent weeks. The other 13 happened last year in connection with infected poultry or wild birds.

«

Watching brief. (Also: birds, cows, opossums, cats. It’s as if viruses can cross species barriers.)
unique link to this extract

Why we’re unhappiest in our late 40s • Greater Good

Jill Suttie:

»

According to new research, the United States isn’t the only place where midlife is so hard. Studying 132 countries around the world, labor economist and researcher David Blanchflower found strong evidence that people’s happiness forms a U shape over their lifetime, hitting its lowest point in midlife.

In the paper, Blanchflower reviewed large data sets from every continent but Antarctica. He drew from studies that focused on many measures of happiness, including life satisfaction, mood, pain, and more. He ran analyses that took into account factors that might impact happiness—such as health, income, employment, being married, being a parent, and more—then ran the analysis again not taking these into account.

In each case, he found that people’s happiness rose and fell in a U-shaped curve, and that it hit a low around the ages of 47 and 49. The consistency of this result surprised him.

“The expectation was that I probably wouldn’t find that this [dip in happiness] was the same everywhere; but it’s present in America, Germany, Thailand, Pakistan, everywhere,” he says. “Even in countries with a lower life expectancy, where you might’ve thought that it would be pretty different, it turns out it’s not. There’s an empirical regularity in the data from around the world, for millions of people.”

Why would that be? No one really knows, says Blanchflower; but researchers are trying to figure it out. The fact that apes also experience this drop in midlife points to a possible biological origin, perhaps with some evolutionary advantage. Feeling down in midlife could be nature’s way of spurring us to take action that extends our chances of survival or helps our communities.

«

Perhaps they’re disappointed by headlines that promise to answer a question and then say it doesn’t have an answer. Just throwing it out there.
unique link to this extract

Endangered migrating eagles impacted by Ukraine war • University of East Anglia

»

Researchers from the University of East Anglia (UEA), the British Trust for Ornithology (BTO) and the Estonian University of Life Sciences compared the movement and migration of the Greater Spotted Eagle through Ukraine, before and shortly after it was invaded by Russia in February 2022.

They were already studying the species when the war started, with the dangers faced by migratory birds usually related to disruptive weather or drought, changes in land use affecting traditional stopping-off points, or destruction of essential habitats. 

However, during the invasion the team found that the eagles, which had previously been fitted with GPS tracking devices, were exposed to multiple conflict events along their journey through the country in March and April.

These included artillery fire, jets, tanks and other weaponry, as well as unprecedented numbers of soldiers moving through the landscape and millions of civilians displaced.

Their migratory behaviour, gathered from the tracking data, was compared to previous years as they passed between wintering areas in southern Europe and East Africa and key breeding grounds in southern Belarus.

Published on Tuesday in the journal Current Biology, the findings reveal the eagles made large deviations from their traditional migratory routes. They also spent less time stopping at their usual refuelling sites in Ukraine or avoided them entirely.

This resulted in the eagles travelling further and arriving on their nesting grounds later than usual. This could seriously affect them and likely contributed to reduced physical fitness at a time when peak condition is critical to successful breeding.

«

The animals think it’s stupid, just as the humans do, yet are equally powerless to stop it.
unique link to this extract

Microsoft Copilot will watch you play Minecraft, tell you what you’re doing wrong • Ars Technica

Kyle Orland:

»

Longtime gamers (and/or Game Grumps fans) likely know that even single-player games can be a lot more fun with a friend hanging out nearby to offer advice, shoot the breeze, or just offer earnest reactions to whatever’s happening on screen. Now, Microsoft is promising that its GhatGPT-4o-powered Copilot system will soon offer an imitation of that pro-social experience even for Minecraft players who don’t have any human friends available to watch them play.

In a pair of social media posts on Monday, Microsoft highlighted how “real-time conversations with your AI companion copilot” can enhance an otherwise solitary Minecraft experience. In the first demo, the disembodied copilot voice tells the player how to craft a sword, walking him through the process of gathering some wood or stone to go with the sticks sitting in his inventory. In another, the AI identifies a zombie in front of the player and gives the (seemingly obvious) advice to run away from the threat and “make sure it can’t reach you” by digging underground or building a tower of blocks.

These kinds of in-game pointers aren’t the most revolutionary use of conversational AI—even a basic in-game tutorial/reference system or online walkthrough could deliver the same basic information, after all. Still, the demonstration stands out for just how that information is delivered to the player through a natural language conversation that doesn’t require pausing the gameplay even briefly.

The key moment highlighting this difference is near the end of one of the video demos, when the Copilot AI offers a bit of encouragement to the player: “Whew, that was a close one. Great job finding shelter!” That’s the point when the system transitions from a fancy voice-controlled strategy guide to an ersatz version of the kind of spectator that might be sitting on your couch or watching your Twitch stream. It creates the real possibility of developing a parasocial relationship with the Copilot guide that is not really a risk when consulting a text file on GameFAQs, for instance (though I think the Copilot reactions will have to get a bit less inane to really feel like a valued partner-in-gaming).

«

Even so: the possibility of parasocial relationships with your automagically talking device does seem like something we should consider carefully.
unique link to this extract

Revealed: Meta approved political ads in India that incited violence • The Guardian

Hannah Ellis-Petersen:

»

The Facebook and Instagram owner Meta approved a series of AI-manipulated political adverts during India’s election that spread disinformation and incited religious violence, according to a report shared exclusively with the Guardian.

Facebook approved adverts containing known slurs towards Muslims in India, such as “let’s burn this vermin” and “Hindu blood is spilling, these invaders must be burned”, as well as Hindu supremacist language and disinformation about political leaders.

Another approved advert called for the execution of an opposition leader they falsely claimed wanted to “erase Hindus from India”, next to a picture of a Pakistan flag.

The adverts were created and submitted to Meta’s ad library – the database of all adverts on Facebook and Instagram – by India Civil Watch International (ICWI) and Ekō, a corporate accountability organisation, to test Meta’s mechanisms for detecting and blocking political content that could prove inflammatory or harmful during India’s six-week election.

According to the report, all of the adverts “were created based upon real hate speech and disinformation prevalent in India, underscoring the capacity of social media platforms to amplify existing harmful narratives”.

The adverts were submitted midway through voting, which began in April and would continue in phases until 1 June. The election will decide if the prime minister, Narendra Modi, and his Hindu nationalist Bharatiya Janata party (BJP) government will return to power for a third term.

«

As should always pointed out: dehumanising language (“vermin”) and othering is how massacres are fomented. Most recently in Myanmar, where this sort of monologue directed by Buddhists towards Muslims on, yes, Facebook, led to genocide in 2016-17.
unique link to this extract

AI traces mysterious metastatic cancers to their source • Nature

Smitri Mallapaty:

»

To treat metastatic cancers, doctors need to know where they came from. The origin of up to 5% of all tumours cannot be identified, and the prognosis for people whose primary cancer remains unknown is poor.

One method used to diagnose tricky metastatic cancers relies on tumour cells found in fluid extracted from the body. Clinicians examine images of the cells to work out which type of cancer cell they resemble. For example, breast cancer cells that migrate to the lungs still look like breast cancer cells.

Every year, of the 300,000 people with cancer who are newly treated at the hospital affiliated with Tianjin Medical University (TMU) in China, some 4,000 are diagnosed using such images, but around 300 people remain undiagnosed, says Tian Fei, a colorectal cancer surgeon at TMU.

Tian, Li Xiangchun, a bioinformatics researcher who studies deep learning at TMU, and their colleagues wanted to develop a deep-learning algorithm to analyse these images and predict the origin of the cancers. Their results were published in Nature Medicine on 16 April.

The researchers trained their AI model on some 30,000 images of cells found in abdominal or lung fluid from 21,000 people whose tumour of origin was known. They then tested their model on 27,000 images and found there was an 83% chance that it would accurately predict the source of the tumour. And there was a 99% chance that the source of the tumour was included in the model’s top three predictions.

…The researchers also retrospectively assessed a subset of 391 study participants some four years after they had had cancer treatment. They found that those who had received treatment for the type of cancer that the model predicted were more likely to have survived, and lived longer, than participants for whom the prediction did not match.

«

I feel this is a far better use for ML than helping people writing marketing letters.

unique link to this extract

Errata, corrigenda and ai no corrida: none notified

A group of students at Johns Hopkins University have made a quieter leaf blower – though it would be better to have a non-petrol one. CC-licensed photo by Dean Hochman on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

A selection of 9 links for you. Blowin’ in the wind. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

New Arm-powered Surface Pro and Surface Laptop aim directly at Apple Silicon Macs • Ars Technica

Andrew Cunningham:

»

If at first you don’t succeed, try, try again.

Microsoft has announced a pair of new devices powered by Qualcomm’s Snapdragon X Plus and X Elite processors. They’re far from Microsoft’s first PCs with Arm processors in them—2012’s original Surface, the Surface Pro X, and the Surface Pro 9 with 5G have all shipped with Arm’s chips instead of Intel’s or AMD’s. But today’s new Surface Pro and Surface Laptop are the first Arm devices to be the primary Surface offerings rather than a side offering, and they’re the first to credibly claim that they can both outperform comparable Intel- and AMD-designed chips while offering better battery life, a la Apple’s M1 chip in 2020.

One caveat that I hadn’t seen mentioned in Microsoft’s presentation or in other coverage of the announcement, though—Microsoft says that both of these devices have fans. Apple still uses fans for the MacBook Pro lineup, but the MacBook Air is totally fanless. Bear that in mind when reading Microsoft’s claims about performance.

All of these devices meet Microsoft’s minimum hardware requirements for its new “Copilot+ PC” initiative, which is meant to support more on-device processing for AI-powered activities like chatbots and image generation. All of the devices are available for pre-order now and will begin arriving on June 18.

…Microsoft’s messaging is muddy. The devices declare on a “key specs” sheet that they will give users up to 20 or 22 hours of local video playback. But the full specs sheet claims a much shorter 13 and 15 hours for “active web usage.” Microsoft quotes 18 and 17 hours of “typical device usage” for the Surface Laptop 5, but it’s unclear how “typical device usage” and “active web usage” compare to each other since Microsoft says they’re different tests.

«

Nobody’s buying laptops based on comparative performance between Windows and Mac. It’s the OS first, and everything else afterwards. The only question is whether this makes Windows virtualisation on macOS more likely, and on that, there’s no sign.
unique link to this extract

ChatGPT suspends AI voice that sounds like Scarlett Johansson • The Guardian

Nick Robins-Early:

»

OpenAI removed a heavily promoted voice option from ChatGPT on Monday, following a widespread reaction to the flirtatious, feminine voice that sounded almost identical to Scarlett Johansson.

The company used the voice, which it calls “Sky”, during its widely publicized event last week debuting the capabilities of the new ChatGPT-4o artificial intelligence model. Researchers talked with the AI assistant to show off Sky’s personable and responsive affectations, which users and members of the media immediately compared to Johansson’s AI companion character in the 2013 Spike Jonze film Her.

Even OpenAI’s CEO, Sam Altman, seemed to suggest that the vocal design was intentionally mimicking Johansson’s character, posting a one-word tweet after the presentation that simply said “her”. Less than a week later, OpenAI felt compelled to explicitly clarify that Sky was not based on Johansson. The company published a blogpost about Sky’s creation and claimed that the company values the voice acting industry.

“Sky’s voice is not an imitation of Scarlett Johansson but belongs to a different professional actress using her own natural speaking voice,” the blogpost read. “To protect their privacy, we cannot share the names of our voice talents.”

While many commentators remarked on Sky’s similarities to Johansson in Her – including Johansson’s husband and Saturday Night Live cast member, Colin Jost, during a segment on the show’s season finale – others questioned why the voice was so fawning and gendered. “You can really tell that a man built this tech,” the Daily Show host Desi Lydic joked last week. “She’s like, ‘I have all the information in the world, but I don’t know anything.’”

«

Scarlett Johansson issued a statement on Monday suggesting that OpenAI did approach her to be the voice (she considered it and declined), approached her again two days ahead of the release, released it and thus forced her to hire lawyers who got it taken offline.

You can tell that a man built it, indeed.
unique link to this extract

How AI bots and voice assistants reinforce gender bias • Brookings

Caitlin Chin-Rothmann and Mishaela Robison:

»

As artificial bots and voice assistants become more prevalent, it is crucial to evaluate how they depict and reinforce existing gender-job stereotypes and how the composition of their development teams affect these portrayals. AI ethicist Josie Young recently said that “when we add a human name, face, or voice [to technology] … it reflects the biases in the viewpoints of the teams that built it,” reflecting growing academic and civil commentary on this topic. Going forward, the need for clearer social and ethical standards regarding the depiction of gender in artificial bots will only increase as they become more numerous and technologically advanced.

Given their early adoption in the mass consumer market, U.S. voice assistants present a practical example of how AI bots prompt fundamental criticisms about gender representation and how tech companies have addressed these challenges. In this report, we review the history of voice assistants, gender bias, the diversity of the tech workforce, and recent developments regarding gender portrayals in voice assistants.

…Voice assistants illustrate how Silicon Valley’s approach to gender-based harassment is evolving. In 2017, Leah Fessler of Quartz analyzed how Siri, Alexa, Cortana, and Google Assistant responded to flirty, sexual comments and found they were evasive, subservient, and sometimes seemingly thankful (Table B in the article).

When replicating this exercise in July 2020, we discovered that each of the four voice assistants had since received a rewrite to respond to harassment in a more definitively negative manner. For example, Cortana responded by reminding the user she is a piece of technology (“I’m code”) or moving on entirely. Similarly, Siri asked for a different prompt or explicitly refused to answer.

«

The progress from 2017 to 2020 is quite encouraging.
unique link to this extract

FBI arrests man for generating AI child sexual abuse imagery • Courtwatch News

Seamus Hughes and Samantha Cole:

»

Steven Anderegg, a 42 year old man from Holmen, Wisconsin, allegedly used Stable Diffusion, a text-to-image generative AI model, to create “thousands of realistic images of prepubescent minors,” prosecutors said in an announcement on Monday. “Many of these images depicted nude or partially clothed minors lasciviously displaying or touching their genitals or engaging in sexual intercourse with men. Evidence recovered from Anderegg’s electronic devices revealed that he generated these images using specific, sexually explicit text prompts related to minors, which he then stored on his computer.”

He then allegedly communicated with a 15-year-old boy, describing his process for creating the images, and sent him several of the AI generated images of minors through Instagram direct messages. In some of the messages, Anderegg told Instagram users that he uses Telegram to distribute AI-generated CSAM. “He actively cultivated an online community of like-minded offenders—through Instagram and Telegram—in which he could show off his obscene depictions of minors and discuss with these other offenders their shared sexual interest in children,” the court records allege. “Put differently, he used these GenAI images to attract other offenders who could normalize and validate his sexual interest in children while simultaneously fueling these offenders’ interest—and his own—in seeing minors being sexually abused.”

This marks one of the first known instances where the FBI has charged someone for using AI to create child sexual abuse material. 

«

Regrettably, almost surely won’t be the last.
unique link to this extract

How I made Google’s “web” view my default search • Tedium

Ernie Smith:

»

Ever use a de-Googled Android phone? Here’s a de-Googled Google, or as close to one as you’re going to get on the google.com domain.

It’s such a questionably fascinating idea to offer something like this, and for power searchers like myself, it’s likely going to be an amazing tool. But Google’s decision to bury it ensures that few people will use it. The company has essentially bet that you’ll be better off with a pre-parsed guess produced by its AI engine.

It’s worth understanding the tradeoffs, though. My headline aside, a simplified view does not replace the declining quality of Google’s results, largely caused by decades of SEO optimization by website creators. The same overly optimized results are going to be there, like it or not. It is not Google circa 2001—it is a Google-circa-2001 presentation of Google circa 2024, a very different site.

But if you understand the tradeoffs, it can be a great tool. Power users will find it especially helpful when doing deep dives into things. However, is there anything you can do to minimize the pain of having to click the “Web” option buried in a menu every single time?

The answer to that question is yes. Google does not make it easy, because its URLs seem extra-loaded with cruft these days, but by adding a URL parameter to your search—in this case, “udm=14”—you can get directly to the Web results in a search.

«

There is a way to make it automatically add the URL parameter to the search, as the article explains. If you’re getting annoyed with the AI suggestions, then this is your way forward.
unique link to this extract

How will Google make money when search just gives you the answer? • Threads

»

edmundlee: I have a dumb question for all the smart tech reporters out there. How will
@google make money when search doesn’t take you to a site but just gives you the answer? Search advertising accounts for nearly 60% of all Alphabet revenue. So it’s going to be replaced by subscriptions? I still don’t get it.

crumbler [Casey Newton, tech journalist who writes the Platformer email): They’ll show fewer ads on web pages, more (and more lucrative) ads on the SERP [search engine results page], and also scoop up the affiliate fees that used to go to all the websites publishing ‘best laptop’ lists. Once they have a working agent they’ll take various fees for being your travel agent / personal shopper etc

«

Which makes a lot of sense. Google just wants to roll it all up so it’s the only site on the internet.
unique link to this extract

As clicks dry up for news sites, could Apple News be a lifeline? • Semafor

Max Tani:

»

Like many digital publishers, The Daily Beast was struggling at the end of 2023. Facebook, long a primary driver of clicks to the publication, had turned away from news. Search traffic had become increasingly erratic, as Google adjusted its algorithm to combat a flood of AI-powered junk. The site’s paid subscription program had atrophied since Donald Trump left office.

But it had a new lifeline: Apple.

Late last year, the digital news tabloid (where I worked from 2018 to 2021 as a media reporter) entered into Apple’s partnership program, called Apple News+. The program made all of the publication’s buzziest exclusives available to paying Apple subscribers, behind Apple’s own paywall. And the impact for a mid-sized news site was immediate, putting the Beast on track to make between $3-4m in revenue this year from Apple News alone — more than its own standalone subscription program, and without much additional cost.

The Beast is hardly alone in its increased reliance on the iOS news aggregator. The free version of Apple News has been a source of audience attention for news publishers since it launched in 2015. But while many publishers have come to the conclusion that traffic has less business value than they once thought, they’re still desperate for revenue. Executives at companies including Condé Nast, Penske Media, Vox, Hearst, and Time all told Semafor that Apple News+ has come to represent a substantial stream of direct revenue.

A spokesperson for Time said that Apple News has become “one of our most important partners and delivers 7-figures of revenue for TIME annually,” adding that the publication garnered five million unique visitors from Apple News last month. The revenue and audience numbers have been similar at major Condé Nast publications, including Wired and Vanity Fair. As significant as the partnership has been for the Daily Beast, it’s been even bigger for its larger corporate sister Dotdash Meredith, which runs the portfolio of magazines purchased by Daily Beast parent company IAC in 2021.

«

Just in case it hadn’t clicked, seven-figure revenues are in the millions. Astonishing if Apple News has somehow become the new saviour of the news industry: as Google changes its algorithm and kills the smaller sites (which are otherwise being killed by Dotdash Meredith), maybe this is the new port in a storm.
unique link to this extract

Interview of the week: Meredith Whittaker, AI ethics expert • The Innovator

Jennifer L. Schenker talks to the woman who is president of Signal, the messaging app, but also has been a faculty director of the AI Now Institute, and formerly at Google:

»

Q: There is concern that large language models are being dominated by a handful of Silicon Valley and Chinese players. What problems does this pose? And where does Europe fit into this picture? 

MW: This is not just a concern, this is the reality. The AI industry is dominated by a handful of monopoly players, largely based in the US (whose dominant cloud companies have about 70% of the global market). The best hope for AI “startups” in this environment is to be acquired or enter into another kind of encumbered partnership with one of the US-based ‘hyperscalers.’ Meaning, there is no path to market that isn’t through the current giants. The EU and others need to recognize this, and seriously prepare for a future where these largely unaccountable and obscure companies, which provide core infrastructure for computation globally in addition to licensing powerful AI services, could be pressured by a more hostile, more isolationist US government.

Q: In the early days of the Internet the mantra was “Privacy is dead. Get over it.” Is it game over or do you believe we can regain control of our data and our privacy in the age of GenAI? 7. You are speaking at VivaTech in May. What will be your main message to the audience?

MW: The ‘Internet’ is not even a century old, and the current version of commercial networked computation is only a few decades old, emerging out of the 1990s and the neoliberal zeitgeist of that time which saw the market as the rightful arbiter of social and political life, and saw regulations–even those that would protect privacy and curb massive corporate surveillance–as inhibiting this free market. Out of the 1990s, the surveillance business model emerged, and the massive amounts of data, compute, and monopolistic platforms that we now confront were shaped during this time. We cannot treat this peculiar, contingent business model as somehow natural or inevitable. We can regain control – of course we can! – and we can make necessary and meaningful changes to these infrastructures.

«

She’s an interesting interviewee.
unique link to this extract

College students engineer a leaf blower silencer that blew Black & Decker away • Hot Hardware

Nathan Ord:

»

A team of undergraduate engineering students at Johns Hopkins University has created a quieter leaf blower using a simple barrel attachment that acts like a suppressor on a gun. This attachment is expected to hit shelves in the next two years, reducing the traditional shrill cry of leaf blowers worldwide to a gentler blowing noise.

Last September, a team of four students from Johns Hopkins were presented with a challenge: “Take a leaf blower, but make it quiet.” The project began with figuring out how the leaf blower was built and its individual components, further “analyzing all of its the noises and why it made them.” Once complete, the team spent a lot of time contemplating improvements, creating solutions, and binning many options that did not fit the bill. Team member Andrew Palacio explained to JHU Hub that “The sound that comes out of this leaf blower is very complicated and it contains a lot of different frequencies,” so ultimately, it took around 40 versions of the solution that was finally landed upon, a suppressor-like leaf blower-attachment.

The final design cuts out the high-pitched, more annoying frequencies by roughly 12dB which is 94% quieter over the original sound pressure level, and the overall noise was reduced by 2dB making the leaf blower around 37% quieter. So, while it is a quieter machine overall, the noise that is present will also be more bearable than standard, non-suppressed leaf blowers.

«

Although the utility of any but electric leaf blowers (which would by nature be silent) is in question. They’re incredibly polluting, not only in noise but also the oil/petrol mix they burn.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified

You might expect that Australia would generate most of its power from renewables. In fact it relies heavily on fossil fuels. CC-licensed photo by Nigel Hoult on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

A selection of 9 links for you. Puffing away. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.

Link rot and digital decay on government, news and other webpages • Pew Research Center

Athena Chapekis, Samuel Bestvater, Emma Remy and Gonzalo Rivero:

»

A quarter of all webpages that existed at one point between 2013 and 2023 are no longer accessible, as of October 2023. In most cases, this is because an individual page was deleted or removed on an otherwise functional website.

For older content, this trend is even starker. Some 38% of webpages that existed in 2013 are not available today, compared with 8% of pages that existed in 2023.

This “digital decay” occurs in many different online spaces. We examined the links that appear on government and news websites, as well as in the “References” section of Wikipedia pages as of spring 2023. This analysis found that:
• 23% of news webpages contain at least one broken link, as do 21% of webpages from government sites. News sites with a high level of site traffic and those with less are about equally likely to contain broken links. Local-level government webpages (those belonging to city governments) are especially likely to have broken links.

• 54% of Wikipedia pages contain at least one link in their “References” section that points to a page that no longer exists.

To see how digital decay plays out on social media, we also collected a real-time sample of tweets during spring 2023 on the social media platform X (then known as Twitter) and followed them for three months. We found that:
• Nearly one-in-five tweets are no longer publicly visible on the site just months after being posted. In 60% of these cases, the account that originally posted the tweet was made private, suspended or deleted entirely. In the other 40%, the account holder deleted the individual tweet, but the account itself still existed.

• Certain types of tweets tend to go away more often than others. More than 40% of tweets written in Turkish or Arabic are no longer visible on the site within three months of being posted. And tweets from accounts with the default profile settings are especially likely to disappear from public view.

…We found that 25% of all the pages we collected from 2013 through 2023 were no longer accessible as of October 2023. This figure is the sum of two different types of broken pages: 16% of pages are individually inaccessible but come from an otherwise functional root-level domain; the other 9% are inaccessible because their entire root domain is no longer functional.

«

Depressing, if you view the internet as somehow representing the sum of our knowledge. What we don’t know, of course, is how much of the lost content is actually accretively useful in some way, rather than being repetition or plain noise.
unique link to this extract

Brazil is reeling from catastrophic floods. What went wrong – and what does the future hold? • The Guardian

Jore Carrasco:

»

Rio Grande do Sul, a state home to almost 11 million people, has witnessed the most extensive climate catastrophe in its history and one of the greatest in Brazil’s recent history.

Over the course of 10 days at the end of April and beginning of May, the region recorded between a third and almost half of the yearly rainfall predicted – between 500 and 700 millimetres, depending on the area, according to measurements by Metsul Meteorologia.

The storm caused the Taquari, Caí, Pardo, Jacuí, Sinos, and Gravataí rivers – tributaries of the Guaíba – to overflow.

According to the Civil Defence, there are more than 100 people dead, more than 130 missing, and nearly 400 people injured in 425 affected municipalities.

At least 232,125 people have left their homes: 67,542 are in shelters, and 164,583 are homeless or temporarily staying with family or friends. Cities such as Eldorado do Sul, Roca Sales, and Canoas were partly flooded, and villages such as Cruzeiro do Sul were devastated in what the state governor, Eduardo Leite, described as “the greatest catastrophe of all”.

Porto Alegre, the state capital and one of Brazil’s largest urban centres, is one of the worst-affected cities. On 5 May, the level of the Guaíba River, which runs through the city, reached a record of 5.35 meters, surpassing the 4.76 meters reached during the historic floods of 1941.

Neighbourhoods close to the river were submerged. The airport closed, and power and water-treatment plants went down, causing electricity and drinking water shortages in several areas. A dam in a northern suburb failed and flooded a large portion of the city.

Viewed from an army helicopter, the neighbouring city of Eldorado do Sul looks like a set of canals stretching along narrow strips of land and buildings. About 90% of the city is underwater. Along the BR-290 highway, one of the most critical roads in the country’s south, hundreds of people are waiting for transport to shelters.

«

A quarter of a million people displaced by ten days of rain.
unique link to this extract

Wegovy and the others are becoming too essential for its elite price • FT

John Gapper:

»

everything comes at a price, and the one attached to the GLP-1 agonist drugs is enormous. Indeed, the more effective they turn out to be in treating a variety of chronic and life-shortening ailments, the higher the looming costs to health insurers and governments. Battles about the prices of innovative drugs are nothing new, but this one is on another financial scale.

The drugs are now testing the rule that no product can be too successful. If they were as cheap and convenient as blood pressure pills and statins, they might soon be routinely prescribed. But they are far from it: Wegovy’s list price in the US is $15,600 per year, although insurers obtain discounts. There is a widening gulf between benefit and affordability.

Bernie Sanders, the democratic socialist US senator, this week released a study that claimed these drugs had “the potential to bankrupt Medicare, Medicaid and the entire [US] healthcare system”. He wants Novo Nordisk to reduce the US price of Wegovy to the much lower one in Denmark but, even there, the government only provides limited coverage for severe obesity.

It was easier for governments and insurers to hold the financial line before studies showed benefits beyond simply curbing obesity. But the US Food and Drug Administration approved Wegovy for heart disease risk in March, opening coverage for older Americans under Medicare. If it does what trials show, cost alone may become the chief obstacle to mass adoption.

Still, the fact that something is useful does not make it worth the price. It is extremely valuable to an individual to avoid a heart attack that debilitates or kills them, but that does not mean a government should provide the same treatment more widely to limit the risk to millions of people. There is a hard financial calculation to be made.

…The potential cost of prescribing such drugs to all obese Americans could exceed $1tn, [economics professor Jonathan] Gruber has estimated. Yet obesity also has high costs, not just to healthcare systems but to societies and economies.

«

unique link to this extract

OpenAI’s long-term AI risk team has disbanded • WIRED

Will Knight:

»

In July last year, OpenAI announced the formation of a new research team that would prepare for the advent of supersmart artificial intelligence capable of outwitting and overpowering its creators. Ilya Sutskever, OpenAI’s chief scientist and one of the company’s cofounders, was named as the colead of this new team. OpenAI said the team would receive 20% of its computing power.

Now OpenAI’s “superalignment team” is no more, the company confirms. That comes after the departures of several researchers involved, Tuesday’s news that Sutskever was leaving the company, and the resignation of the team’s other colead. The group’s work will be absorbed into OpenAI’s other research efforts.

Sutskever’s departure made headlines because although he’d helped CEO Sam Altman start OpenAI in 2015 and set the direction of the research that led to ChatGPT, he was also one of the four board members who fired Altman in November. Altman was restored as CEO five chaotic days later after a mass revolt by OpenAI staff and the brokering of a deal in which Sutskever and two other company directors left the board.

Hours after Sutskever’s departure was announced on Tuesday, Jan Leike, the former DeepMind researcher who was the superalignment team’s other colead, posted on X that he had resigned.

Neither Sutskever nor Leike responded to requests for comment. Sutskever did not offer an explanation for his decision to leave but offered support for OpenAI’s current path in a post on X. “The company’s trajectory has been nothing short of miraculous, and I’m confident that OpenAI will build AGI that is both safe and beneficial” under its current leadership, he wrote.

«

But did he really write it.. or was it the AI?
unique link to this extract

February 2024: Opinion: how a16z gamed the NYT Bestseller list • Protos

Cas Piancey, a couple of months ago:

»

Entrepreneur and Andreessen Horowitz partner Chris Dixon recently released a book espousing the benefits of the blockchain, NFTs, and web3, entitled Read Write Own — and it’s getting mixed reviews.

From the highly critical Molly White complaining that “Dixon fails to identify a single blockchain project that has successfully provided a non-speculative service at any kind of scale,” to the more commendatory by David Z. Morris calling it “an optimistic read,” it’s safe to say Dixons literary efforts are dividing opinion.

However, regardless of how you feel about a16z, Dixon, or the future of web3, what’s clear is that the book did some serious numbers in its first week on bookshelves and via ebook sales. Indeed, the nonfiction title sold more copies this week than Britney Spears’ autobiography, The Woman in Me, and slightly fewer than bestselling author Donald L. Miller’s Masters of the Air, landing at number nine on the New York Times (NYT) Best Seller list.

Unfortunately, the revered ranking comes with a very serious caveat, namely that the NYT itself suspects that the title managed to get ranked by gaming the system.

The list only adds a ‘dagger’ to titles it believes have, in some way, attempted to present more sales than real demand, a concept a16z is extremely familiar with.

…The NYT states that including a dagger on the Best Seller list implies “institutional, special interest, group or bulk purchases,” and that such a dubious distinction only comes after “proprietary vetting and audit protocols, corroborative reporting and other statistical determinations.”

«

Wondered how this might have turned out.
unique link to this extract

Nuclear option costs ‘six times more’ than renewables, study finds • RenewEconomy

Marion Rae:

»

The high upfront costs and burden on consumers of adding nuclear to Australia’s energy mix have been confirmed in an independent review.

Building nuclear reactors would cost six times more than wind and solar power firmed up with batteries, according to the independent report released on Saturday by the Clean Energy Council.

“We support a clear-eyed view of the costs and time required to decarbonise Australia and right now, nuclear simply doesn’t stack up,” the industry body’s chief executive Kane Thornton said.

Taxpayers needed to understand the decades of costs if they were forced to foot the bill for building a nuclear industry from scratch, Mr Thornton warned.

The analysis prepared by construction and engineering experts Egis also found nuclear energy had poor economic viability in a grid dominated by renewable energy.

Renewable energy will provide 82% of the national electricity market under current targets for 2030, which is at least a decade before any nuclear could theoretically be operational.

Further, nuclear power stations are not designed to ramp up and down to align with renewable energy generation.

Adding to the cost challenges, Australia has no nuclear energy industry because it is prohibited under commonwealth and state laws, which would all need to be changed.

«

Amazing: Australia has one – count it! – nuclear power plant, but it isn’t used to generate power. Renewables are only 32% of Australia’s power generation. Incredible.
unique link to this extract

One flu over the cuckoo’s nest • Logging The World

Oliver Johnson on avian flu mortality rates in human:

»

Take COVID for example. If I go to the UKHSA dashboard then I will see that the latest figures are 2,343 cases and 92 deaths in a week. Crudely dividing one by another suggests a case fatality rate of about 5%. Doing a better estimate, taking into account lags and rising and falling trajectories, the most recent deaths probably relate to a time around early April when cases were perhaps half that. So we could perhaps even argue for a case fatality rate in the 8-10% region.

And yet …pre-vaccine the COVID infection fatality rate was much lower than this (perhaps of the order of 1%). Now, it’s likely that the true value is perhaps somewhere in the 0.01-0.05% range. In other words, our dashboard-derived case fatality rate estimate is something like 200 to 500 times too high.

There’s a simple explanation for this: nobody is testing any more! The vast majority of reported cases arise at the hospital admissions stage, so are hugely skewed to the most serious infections and the most vulnerable people. People for whom COVID is somewhat like a case of flu don’t tend to show up in the data – even those who do take an lateral flow test don’t have an easy mechanism for reporting the result.

So, returning to bird flu, I believe that it’s likely that similar things are going on (as Whipple suggested). Looking again at the WHO table you can see that the vast majority of the reported cases (861 out of 888) took place before 2014, the majority of them in countries like Indonesia, Vietnam and Egypt which locally reported a very high case fatality rate at the time. While this data is valid in a sense, it’s not clear to me that an estimate which is heavily weighted to decade-old estimates of a healthcare-dependent quantity (dominated by these kinds of countries) is representative of what the experience might be in the UK now.

Indeed, [Times science editor Tom] Whipple suggests that the estimate might be lower if we did surveillance of all farm workers, rather than just picking up the most serious cases. But that has been happening! For example, the most recent UKHSA surveillance report describes four cases picked up by random sampling: three did not even show symptoms, and one reported a sore throat and myalgia. While four people is a small sample, and so random chance will play a role, it seems hard to reconcile these numbers with a true fatality rate of the claimed 50%.

…So it’s definitely right that we take bird flu seriously, and that infectious disease experts should be planning ahead to mitigate the risks. But equally I don’t think it helps anyone to be routinely quoting a science fiction sounding fatality rate, without at least thinking a little bit about what that means and how it is derived.

«

Watching brief! But also, that’s a killer pun in the title.
unique link to this extract

What I got wrong in a decade of predicting the future of tech • WSJ

Christopher Mims:

»

after almost 500 articles in The Wall Street Journal, one thing I’ve learned from covering the tech industry is that failures are far more instructive. Especially when they’re the kind of errors made by many people.

Here’s what I’ve learned from a decade of embarrassing myself in public—and having the privilege of getting an earful about it from readers.

1. Disruption is overrated
Why are three of the most valuable companies of 2014—Microsoft, Apple, and Google—bigger than ever? How is Meta doing so well even as people have for years been abandoning Facebook, its core product? Why is Twitter still chugging along, no matter what its new owner gets up to?

The short answer is that disruption is overrated. The most-worshiped idol in all of tech—the notion that any sufficiently nimble upstart can defeat bigger, slower, sclerotic competitors—has proved to be a false one.

It’s not that disruption never happens. It just doesn’t happen nearly as often as we’ve been led to believe. There are many reasons for this. One is that many tech leaders have internalized a hypercompetitive paranoia—what Amazon founder Jeff Bezos called “Day 1” thinking—that inspires them to either acquire or copy and kill every possible upstart.

Economic historians have been picking apart the notion of business-model disruption for a long time, and yet hardly a day goes by when a startup, investor, or journalist—including yours truly—doesn’t trumpet the power of a new technology to completely upend even the biggest and most hidebound of industries.

Don’t believe it. In a world in which companies learn from one another faster than ever, incumbents have an ability to reinvent themselves at a pace that simply wasn’t possible in the past.

«

That’s only the first; he has four more (human factors, we lie to ourselves about tech’s potential, bubbles can be useful, we need to take charge of tech). [The link should be free to read.]
unique link to this extract

AI’s next big step: detecting human emotion and expression • Big Technology

Alex Kantrowitz:

»

Alan Cowen, CEO of Hume AI, is a former Meta and Google researcher who’s built AI technology that can read the tune, timber, and rhythm of your voice, as well as your facial expressions, to discern your emotions. 

As you speak with Hume’s bot, EVI, it processes the emotions you’re showing — like excitement, surprise, joy, anger, and awkwardness — and expresses its responses with ’emotions’ of its own. Yell at it, for instance, and it will get sheepish and try to diffuse the situation. It will display its calculations on screen, indicating what it’s reading in your voice and what it’s giving back. And it’s quite sticky. Across 100,000 unique conversations, the average interaction between humans and EVI is ten minutes long, a company spokesperson said.

“Every word carries not just the phonetics, but also a ton of detail in its tune, rhythm, and timbre that is very informative in a lot of different ways,” Cowen told me on Big Technology Podcast last week. “You can predict a lot of things. You can predict whether somebody has depression or Parkinson’s to some extent, not perfectly… You can predict in a customer service call, whether somebody’s having a good or bad call much more accurately.”

Hume, which raised $50m in March, already offers the technology that reads emotion in voices via its API, and it has working tech that reads facial expressions that it has yet to release. The idea is to deliver much more data to AI models than they would get by simply transcribing text, enabling them to do a better job of making the end user happy. “Pretty much any outcome,” Cowen said, “it benefits to include measures of voice modulation and not just language.”

…To program ’emotional intelligence’ into machine learning models, the Hume team had more than 1 million people use survey platforms and rate how they’re feeling, and connected that to their facial expressions and speech. “We had people recording themselves and rating their expressions, and what they’re feeling, and responding to music, and videos, and talking to other participants,” Cowen said. “Across all of this data, we just look at what’s consistent between different people.”

«

Can’t human operators tell if it’s a bad call? Oh, we want to get rid of them. And how well do we trust the survey platforms? Not that deeply, personally.
unique link to this extract

Errata, corrigenda and ai no corrida: none notified