Start Up No.2001: Google expected to show AI at I/O, Florida faces its climate problem, spied on by TikTok, cleansing Twitter, and more

There are signs that Mark Zuckerberg’s excitement about the metaverse is waning in favour of AI. Well, who can blame him? It’s what people want. CC-licensed photo by Steve Jurvetson 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.

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

A selection of 9 links for you. Quite the odyssey. I’m @charlesarthur on Twitter. On Mastodon: Observations and links welcome.

Google IO to feature AI updates, showing off PaLM 2 LLM • CNBC

Jennifer Elias:


Artificial intelligence is going to be a central theme at Google’s annual developer conference on Wednesday, as the company is planning to announce a number of generative AI updates, including launching a general-use large language model (LLM), CNBC has learned.

According to internal documents about Google I/O viewed by CNBC, the company will unveil PaLM 2, its most recent and advanced LLM. PaLM 2 includes more than 100 languages and has been operating under the internal codename “Unified Language Model.” It’s also performed a broad range of coding and math tests as well as creative writing tests and analysis.

At the event, Google will make announcements on the theme of how AI is “helping people reach their full potential,” including “generative experiences” to Bard and Search, the documents show. Pichai will be speaking to a live crowd of developers as he pitches his company’s AI advancements.

The updates come as competition ramps up in the AI arm’s race, with Google and Microsoft racing to incorporate chat AI technology into their products. Microsoft is using its investment in ChatGPT creator OpenAI to bolster its Bing search engine, while Google has quickly mobilized to try and incorporate its Bard technology and its own LLM across various teams.

Google first announced the PaLM language model in April of 2022. In March of this year, the company launched an API for PaLM alongside a number of AI enterprise tools it says will help businesses “generate text, images, code, videos, audio, and more from simple natural language prompts.” 

Last month, Google said its medical LLM called “Med-PaLM 2” can answer medical exam questions at an “expert doctor level” and is accurate 85% of the time.


Remember that sorta-AI doodad which was going to phone your hairdresser and make an appointment for you? Whatever happened to that? Google has a bad habit of announcing things at I/O which don’t appear. Maybe these will be different?
unique link to this extract

May 2001: Scientists warn of more CJD cases • BBC On This Day

May 2001:


Leading experts on new variant CJD, the human form of BSE or “mad cow” disease, have warned the current outbreak could get much worse.

So far, 99 people have had the disease and nearly all of them have died.

New evidence gathered from experiments on mice suggests this first batch of cases could be followed in a few years’ time by a much larger “second wave”.

Professor John Collinge is one of the government’s top advisors on vCJD and director of the Medical Research Council Prion Unit in London.

He has found that a small number of the mice he observed got vCJD fairly quickly while the rest had a longer incubation period before contracting the disease.

“I don’t want to be alarmist about this,” he said “but it’s entirely possible and we have to consider that what we are looking at, at the moment is, thankfully, a very small incidence of the disease amongst a small sub-section of the population. It may be five or ten years before the rest of the population of those at risk develop the disease.”


The government had admitted the existence of vCJD back in 1996, but there was still uncertainty about how many people might develop the disease – an awful descent into helplessness as holes developed in the brain. I discovered a study, which wasn’t published, by the official vCJD unit, which estimated the final death toll would be in the hundreds, perhaps peaking in 2003.

The most recent figures show 178 deaths since the first ones were recorded in 1995, with none from 2017. The peak, of 28 deaths, was in 2000.
unique link to this extract

The Metaverse, Zuckerberg’s tech obession, is officially dead. ChatGPT killed it • Business Insider

Ed Zitron:


The Metaverse fell seriously ill as the economy slowed and the hype around generative AI grew. Microsoft shuttered its virtual-workspace platform AltSpaceVR in January 2023, laid off the 100 members of its “industrial metaverse team,” and made a series of cuts to its HoloLens team. Disney shuttered its Metaverse division in March, and Walmart followed suit by ending its Roblox-based Metaverse projects. The billions of dollars invested and the breathless hype around a half-baked concept led to thousands — if not tens of thousands — of people losing their jobs.

But the Metaverse was officially pulled off life support when it became clear that Zuckerberg and the company that launched the craze had moved on to greener financial pastures. Zuckerberg declared in a March update that Meta’s “single largest investment is advancing AI and building it into every one of our products.” Meta’s chief technology officer, Andrew Bosworth, told CNBC in April that he, along with Mark Zuckerberg and the company’s chief product officer, Chris Cox, were now spending most of their time on AI. The company has even stopped pitching the Metaverse to advertisers, despite spending more than $100 billion in research and development on its mission to be “Metaverse first.” While Zuckerberg may suggest that developing games for the Quest headsets is some sort of investment, the writing is on the wall: Meta is done with the Metaverse.


Zitron is (when he’s not being a PR guy) a polemicist, rather than a journalist. However it’s hard to disagree: everyone’s talking about ChatGPT, nobody’s talking about the metaverse. Got to feel Apple’s going to be facing some serious headwinds if it does release a VR headset. (I’m still sceptical.)
unique link to this extract

I know the motive behind every mass shooting • McSweeney’s Internet Tendency

Kimberly Harrington:


“The shooting came barely a week after a man fatally shot five people in Cleveland, Texas, after a neighbor asked him to stop firing his weapon while a baby slept. It also follows other rampages in recent days, including the fatal shootings of six victims in a home in Oklahoma City on Monday, and gunfire that killed one and injured four in a medical facility in Atlanta on Wednesday.” — NBC News

CLEVELAND, TEXAS: The shooter’s motive was to kill people using a gun.

OKLAHOMA CITY: The shooter’s motive was to kill people using a gun.

ATLANTA: The shooter’s motive was to kill people using a gun.

I’m not a cop, a star witness, an FBI agent, or a forensic psychologist. I’m not a clairvoyant, I’m bad at math, and I lie to my doctor about doing regular breast exams. But even I—an ordinary American citizen with soft morals and the ability to take in basic information and discern simple patterns—know what the motive was behind every mass shooting. Every single one.


Put like this, it is pretty obvious. All the chin-scratching around whether the latest mass shooting was by someone who was an extreme right-winger or just a bit right-wing or what tends to overlook this simple, basic fact: such people exist in other countries too, including the UK. And yet: the mass shootings are a thing you only find in the US. Because…

unique link to this extract

Florida tosses climate lifeline to swamped ‘Keybillies’ • E&E News

Daniel Cusick:


A steady outflow of low- and middle-income residents — beginning after Hurricane Wilma in 2005 and again after 2017’s Hurricane Irma — has demographically reshaped Big Pine Key. Longtime residents who work the service and labour jobs undergirding the Keys’ multibillion-dollar tourism economy are being squeezed out. New people with deeper pockets and greater mobility are moving in, often with cash in hand.

In the years after Wilma, Big Pine Key lost a fourth of its population, bottoming out at 3,777 people in 2012, according to Census Bureau data. It rebounded over the next decade — a relatively quiet period for Florida hurricanes — peaking at 5,339 in 2017. Then came Irma, the Keys’ second-strongest storm in a century. It whittled Big Pine Key’s population back down to 4,521 in three years, a 15% drop.

After each hurricane, Big Pine Key’s low-income residents had little to return to, and those who did found themselves packed into substandard houses or mobile home parks. Some sprawling trailer communities with reputations for colorful tenants and rough-and-ready living shuttered after Irma, drawing complaints of no-notice evictions and money-grubbing landlords.

What a difference five years makes. Big Pine Key is rebuilding again, and real estate values have nearly doubled since Irma, from $390,000 in 2018 to $777,000 in 2022, according to the real estate site Redfin. Cleared lots are going for six figures, and even unbuildable lots bound by state-imposed growth restrictions can fetch $80,000 or more. Those prices are putting added pressure on long-term Keys residents.

[Saima] Kawzinsky and her fiancé of 10 years currently pay $1,650 per month in rent for a 2-bedroom elevated home built by an affordable housing land trust created by a local philanthropist.

“We want to buy a house, but it’s getting harder and harder,” she said. “At the end of the day, it’s either going to be me and my family living an underwater lifestyle or leaving and getting our heads just above water.”


“Her fiancé of 10 years” is a phrase that tells its own story. Of course climate change is affecting the low-paid in America first; that’s how it’s going to go everywhere.
unique link to this extract

TikTok spied on me. Why? • FT via Ars Technica

Cirstina Criddle:


One evening in late December last year, I received a cryptic phone call from a PR director at TikTok, the popular social media app. I’d written extensively about the company for the Financial Times, so we’d spoken before. But it was puzzling to hear from her just before the holidays, especially since I wasn’t working on anything related to the company at the time.

The call lasted less than a minute. She wanted me to know, “as a courtesy,” that The New York Times had just published a story I ought to read. Confused by this unusual bespoke news alert, I asked why. But all she said was that it concerned an inquiry at ByteDance, TikTok’s Chinese parent company, and that I should call her back once I’d read it.

The story claimed ByteDance employees accessed two reporters’ data through their TikTok accounts. Personal information, including their physical locations, had been used as part of an attempt to find the writers’ sources, after a series of damaging stories about ByteDance. According to the report, two employees in China and two in the US left the company following an internal investigation. In a staff memo, ByteDance’s chief executive lamented the incident as the “misconduct of a few individuals.”

When I phoned the PR director back, she confirmed I was one of the journalists who had been surveilled. I put down my phone and wondered what it meant that a company I reported on had gone to such lengths to restrict my ability to do so. Over the following months, the episode became just one in a long series of scandals and crises that call into question what TikTok really is, and whether the company has the world-dominating future that once seemed inevitable.


Even the FT’s security team couldn’t stop TikTok trying to invade her privacy.
unique link to this extract

The new Microsoft Bing AI wants to be the future of everything • Fast Company

Ryan Broderick:


The new version of Bing runs on OpenAI’s newest language model, GPT-4. It integrates it further into the Bing search engine and the Edge web browser and centralizes all of its various processes into one interface—including the ability to generate images right in the same window, using OpenAI’s DALL-E 2 image tool. But Bing AI doesn’t just pull all of this into one window, it offers some significant upgrades on how these services work.

The big addition to Bing is chat history. If you’ve never used OpenAI’s ChatGPT, the AI creates a thread for every query you ask. If you come back to that thread, the AI just picks up where you left off. Bing now has that feature, as well, but it also allows you to name your chat threads, organize them, and export them to a PDF or Microsoft Word document.

Bing also has real-time access to Bing search results, a feature that, in its early stages, led to critics describing it as “psychotic.” The chatbot seems a lot more mild-mannered these days, and as a guard against fake results (aka hallucinations), when it returns summaries, it does so with citations pulled in from Bing search. Even more impressive, when running Bing’s AI inside the Edge browser, it can summarize long articles and will include in-page citations, allowing you to find the exact text it’s referencing.

I asked Bing for advice about where to go while I’m on a trip in Milan next week, and it spit out five decent-enough suggestions. It also allowed me to click on the links it was summarizing and then, via an AI chatbot sidebar in the Edge browser, it was able to answer specific questions about the web page I was viewing.

Sarah Mody, Microsoft’s director of global search and AI product marketing, tells Fast Company that they’re hoping to inspire “magic moments,” where the AI suddenly surprises you with what it’s capable of. For her, it was a realization that Bing AI could not only generate a recipe, but also organize the ingredients based on where they would likely be found in the supermarket (an admittedly impressive feat).


OK, it’s impressive – for about ten seconds. But you have to go around the supermarket. Sure, vegetables will be by the entrance and drinks nearer the exit, but don’t you know, Bing, that this stuff gets shuffled around all the time?

Anyhow, the tsunami continues.
unique link to this extract

Your Twitter feed sucks now. These free add-ons can help • WIRED

Justin Pot:


I’m not here to judge anyone who pays for Twitter Blue. However, I’m not a fan of pay-to-win mechanics. At this point, Twitter is a game where players compete for the most attention; Twitter Blue is overpowered DLC [downloadable content]. If you buy a subscription, your tweets are shown at the top of comment threads and prioritized in other contexts, including the “For You” page. This makes Blue the social media equivalent of paying for unlimited ammo or improved body armour, regardless of who you are or whether what you have to say is worth promoting.

It also makes Twitter really annoying to use. No one wants to play with the people who are paying to win. That’s why there’s so much mockery of, and desire to avoid, Twitter Blue users.

Granted, there are plenty of reasons why longtime Twitter power users—in particular, public figures or people who historically have trouble with impersonators—might pay for Twitter Blue. No one wants to lose an audience they worked hard to build, and Twitter has the right to monetize itself however it likes. You also have the right to consume (or not consume) the social media you choose.

All the same, it can’t be denied that a lot of really annoying posts from Twitter Blue users with low follower counts show up in all kinds of contexts. If you don’t want to see those posts, I don’t blame you, and here’s how you can filter them all out.


Basically, use a Chrome extension. Or just look at the people you’re Following. (Though I suspect that some of the people who you follow are shifted to the For You tab, meaning you miss them. Can’t prove it, though.)
unique link to this extract

Marissa Mayer on AI, tech fears, and Yahoo regrets • Morning Brew

Patrick Kulp speaks to the ex-CEO of Yahoo, ex-veep of Google, who is now running an AI startup that wants to organise your contacts list:


PK: If there is one thing that you could have done differently in your time at Yahoo, what would it be?

MM: My perspective on Yahoo is there are probably three things I would have done differently. One is obvious—I hired the wrong COO; I would have hired a different COO. I would have hired [current Integral Ad Science CEO] Lisa Utzschneider, who became my chief revenue officer. And that would have been great.

We looked at a transformative acquisition, and we bought Tumblr [for $1.1bn]. At the same time, we were also considering whether it was possible to buy Hulu or, ironically, Netflix. And I think Netflix was $4bn and Hulu was at $1.3bn at the time. And either of those, with hindsight being 20/20, would have been a better acquisition.

And probably the biggest one—if you made me name just one—is that we should have done the tax-free Alibaba spinoff to separate the assets of the company. Because one, if we had done that, it would have saved $10bn for our shareholders or made them that money, whichever way you look at it, in taxes that were paid. And two, it would have allowed Yahoo to continue as an independent company, and it would have potentially had more success. Now it is an independent company and privately held by private equity. But I’m not sure that the foray through Verizon was as helpful to some of the technologies and what they had to offer as it could have been.


Of course you’re wondering who the “wrong COO” was. It was Henrique de Castro, who she hired over from Google but binned after a year (to January 2014)

Wonder if Netflix would have been for sale. It would have been a brutal takeover fight, and probably would have flamed out over the huge cashflow problem: Mayer would have wanted to data it to death, but Netflix is a content company, not a data company. So, bullet dodged.
unique link to this extract

• Why do social networks drive us a little mad?
• Why does angry content seem to dominate what we see?
• How much of a role do algorithms play in affecting what we see and do online?
• What can we do about it?
• Did Facebook have any inkling of what was coming in Myanmar in 2016?

Read Social Warming, my latest book, and find answers – and more.

Errata, corrigenda and ai no corrida: none notified

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.