Start Up No.2383: the talkative AI future, Johansson demands deepfake law, AI summaries no good for news, and more


The introduction of America’s first ever congestion pricing scheme in New York has been a big success in just one month. CC-licensed photo by adrian8_8 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. Transported. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.


AI finds its voice: we are now entering the conversational computing era • Daniel Oran

Daniel Oran:

»

many current uses of AI involve handing off tasks like drafting an email or writing computer code.

This might be called the delegation model of AI: it works for us, like an eager-to-please employee. You give instructions, then you can walk away — like the “batch mode” of old mainframe computers, which processed stacks of punch cards while you went for coffee.

But some of the best-known early speculation about the future of computing — from Vannevar Bush’s “As We May Think” in 1945 to Douglas Engelbart’s “Augmenting Human Intellect” in 1962 — focused instead on collaboration: imagining how smart machines might work with us to enhance our abilities.

They thought that we’d interact directly with machine intelligence, instead of just delegating and walking away. Interacting right now instead of getting results later makes the experience very different.

In 1987, Apple released a speculative (and prophetic) video, “Knowledge Navigator,” with a similarly collaborative conception of AI. A college professor speaks with an AI assistant while using a touchscreen computer. The interaction feels like a give-and-take with a graduate student.

We’re heading toward that kind of collaborative voice AI, but it’s still early days. As yet, no real-world product comes close to the fictional Knowledge Navigator. It’s probably going to take a while, and if history is any guide, the breakthroughs may not come from today’s dominant players.

In the early ’90s, as the graphical user interface became increasingly popular, software leaders like Lotus and WordPerfect withered as they struggled to make the leap. And after the Apple iPhone changed the interaction paradigm in 2007, phone giants like BlackBerry and Nokia tanked. Conversational computers may turn out to be a similar watershed.

«

It’s worth looking at Knowledge Navigator again, because once you get past the Jeeves-style avatar a lot of it is starting to come into sight.
unique link to this extract


Scarlett Johansson urges limits on A.I. after viral video • People

Lawrence Yee:

»

Scarlett Johansson is urging U.S. legislators to place limits on artificial intelligence as an unauthorized, A.I.-generated video of her and other Jewish celebrities opposing Kanye West goes viral.

The video, which has been circulating on social media, opens with an A.I. version of Johansson, 40, wearing a white T-shirt featuring a hand and its middle finger extended. In the center of the hand is a Star of David. The name “Kanye” is written underneath the hand.

The video contains A.I.-generated versions of over a dozen other Jewish celebrities, including Drake, Jerry Seinfeld, Steven Spielberg, Mark Zuckerberg, Jack Black, Mila Kunis and Lenny Kravitz. It ends with an A.I. Adam Sandler flipping his finger at the camera as the Jewish folk song “Hava Nagila” plays.

The video ends with “Enough is Enough” and “Join the Fight Against Antisemitism.”

In a statement to PEOPLE, Johansson denounced what she called “the misuse of A.I., no matter what its messaging.”

She continued, “It has been brought to my attention by family members and friends, that an A.I.-generated video featuring my likeness, in response to an antisemitic view, has been circulating online and gaining traction. I am a Jewish woman who has no tolerance for antisemitism or hate speech of any kind. But I also firmly believe that the potential for hate speech multiplied by A.I. is a far greater threat than any one person who takes accountability for it. We must call out the misuse of A.I., no matter its messaging, or we risk losing a hold on reality.”

…[She added] “There is a 1000-foot wave coming regarding A.I. that several progressive countries, not including the United States, have responded to in a responsible manner. It is terrifying that the U.S. government is paralyzed when it comes to passing legislation that protects all of its citizens against the imminent dangers of A.I.”

«

I’m not optimistic that the US will pass sensible deepfake legislation, but Johansson might be the best chance for it to happen.
unique link to this extract


A million cars have disappeared: what NYC is like after one month of congestion pricing • Fast Company

Kristin Toussaint:

»

New York City’s congestion pricing program has been in place for one month, implementing tolls on drivers who enter certain, often gridlocked, areas of Manhattan. And so far, the results are “undeniably positive,” transit officials say, with measurably reduced traffic and more commuters choosing public transit. 

The traffic mitigation plan covers a “congestion relief zone” that spans almost all of Manhattan below 60th street and includes major routes like the Lincoln, Holland, and Hugh L. Carey Tunnels and bridges that go into both Brooklyn and Queens. Since its launch on January 5, one million fewer vehicles have entered that zone than they would have without the toll, according to the Metropolitan Transportation Authority (MTA).

Passenger cars with an E-ZPass that travel through that zone face a $9 toll during peak hours, from 5 a.m. to 9 p.m. on weekdays and 9 a.m. to 9 p.m. on weekends, and a $2.25 toll overnight. Tolls are more expensive for commercial traffic, and vehicles without E-ZPass face a 50% premium.

Those charges are meant to reduce traffic in the city and also raise funds for $15bn worth of transit repairs to the MTA. By cutting traffic and ushering more commuters onto public transit, the program will also reduce air pollution and greenhouse gas emissions.

It’s the first such plan in the United States, though congestion pricing has been successfully used in London, Stockholm, Singapore, and other cities. In Stockholm, traffic levels dropped about 25%, and the city saw less pollution and more investment in local infrastructure.

…More commuters are opting for buses to cross Manhattan, and those buses are now traveling more quickly, too. Weekday bus ridership has grown 6%, while weekend ridership is up 21%, compared to January 2024. Subway ridership has also grown by 7.3% on weekdays and 12% on weekends, part of a larger trend in ridership growth happening since the fall, per the MTA.

«

The first! Congestion! Pricing! Plan! In the United States! It’s just incredible that it has taken them this long, and required traffic to get this bad, before they did this. Car transit times are faster too.

Next step – but how soon? – would be a Low Emission Zone, and then a ULEZ which rewards electric cars.
unique link to this extract


Thomson Reuters wins first major AI copyright case in the US • WIRED

Kate Knibbs:

»

Thomson Reuters has won the first major AI copyright case in the United States.

In 2020, the media and technology conglomerate filed an unprecedented AI copyright lawsuit against the legal AI startup Ross Intelligence. In the complaint, Thomson Reuters claimed the AI firm reproduced materials from its legal research firm Westlaw. Today, a judge ruled in Thomson Reuters’ favor, finding that the company’s copyright was indeed infringed by Ross Intelligence’s actions.

“None of Ross’s possible defenses holds water. I reject them all,” wrote US District Court of Delaware judge Stephanos Bibas, in a summary judgement.

Thomson Reuters and Ross Intelligence did not immediately respond to requests for comment.

The generative AI boom has led to a spate of additional legal fights about how AI companies can use copyrighted material, as many major AI tools were developed by training on copyrighted works including books, films, visual artwork, and websites. Right now, there are several dozen lawsuits currently winding through the US court system, as well as international challenges in China, Canada, the UK, and other countries.

«

Ben Thompson points out (in a paywalled post at Stratechery) that this is not as big as it’s being made out to be. Ross Intelligence was using specific Reuters data which it had been forbidden to use. But he thinks the verdict probably won’t survive appeal (even though Ross Intelligence has gone bust, an insurance company is funding the case and may choose to double down rather than quit): “at least two other cases held that there was no copyright violation as long as the copyright material did not appear in the final output.”
unique link to this extract


AI summaries turn real news into nonsense, BBC finds • The Register

Richard Currie:

»

Still smarting from Apple Intelligence butchering a headline, the BBC has published research into how accurately AI assistants summarize news – and the results don’t make for happy reading.

In January, Apple’s on-device AI service generated a headline of a BBC news story that appeared on iPhones claiming that Luigi Mangione, a man arrested over the murder of healthcare insurance CEO Brian Thomson, had shot himself. This was not true and the public broadcaster complained to the tech giant.

Apple first promised software changes to “further clarify” when the displayed content is a summary provided by Apple Intelligence, then later temporarily disabled News and Entertainment summaries. It is still not active as of iOS 18.3, released in the last week of January.

But Apple Intelligence is far from the only generative AI service capable of news summaries, and the episode has clearly given the BBC pause for thought. In original research [PDF] published on Tuesday, Pete Archer, Programme Director for Generative AI, wrote about the corporation’s enthusiasm for the technology, detailing some of the ways in which the BBC had implemented it internally, from using it to generate subtitles for audio content to translating articles into different languages.

“AI will bring real value when it’s used responsibly,” he said, but warned: “AI also brings significant challenges for audiences, and the UK’s information ecosystem.”

The research focused on OpenAI’s ChatGPT, Microsoft’s Copilot, Google’s Gemini, and Perplexity assistants, assessing their ability to provide “accurate responses to questions about the news; and if their answers faithfully represented BBC news stories used as sources.”

«

Spoiler: they’re not good at representing it.
unique link to this extract


Getting rid of the penny introduces a new problem: nickels • CNN Business

Chris Isidore:

»

President Donald Trump says he has ordered the US Mint to stop making pennies, which he correctly says cost more than one cent to produce.

“For far too long the United States has minted pennies which literally cost us more than 2 cents. This is so wasteful!” he said in a post on his Truth Social platform. “Let’s rip the waste out of our great nations budget, even if it’s a penny at a time.”

Trump actually undersold the cost argument — pennies cost more than 3 cents to produce.

But there’s a problem with his plan: phasing out the penny could result in needing to make more nickels, and the US Treasury Department loses far more money on every nickel than it does on every penny.

“Without the penny, the volume of nickels in circulation would have to rise to fill the gap in small-value transactions. Far from saving money, eliminating the penny shifts and amplifies the financial burden,” said American for Common Cents, a pro-penny group funded primarily by Artazn, the company that has the contract to provide the blanks used to make pennies.

According to the latest annual report from the US Mint, each penny cost 3.7 cents to make, including the 3 cents for production costs, and 0.7 cents per coin for administrative and distribution costs. But each nickel costs 13.8 cents, with 11 cents of production costs and 2.8 cents of administrative and distribution costs.

«

It does show how messed up the US has become that it’s losing money on making money, and that by trying to save money on making money it’s going to lose even more money. The fun will be in how long pennies will remain legal tender, and whether people will start melting them down or selling them in bulk instead.
unique link to this extract


Voters were right about the economy. The data was wrong • POLITICO

Eugene Ludwig:

»

Having served as comptroller of the currency during the 1990s, I‘ve spent substantial chunks of my career exploring the gaps between public perception and economic reality, particularly in the realm of finance. Many of the officials I’ve befriended and advised over the last quarter-century — members of the Federal Reserve, those running regulatory agencies, many leaders in Congress — have told me they consider it their responsibility to set public opinion aside and deal with the economy as it exists by the hard numbers. For them, government statistics are thought to be as reliable as solid facts.

In recent years, however, as my focus has broadened beyond finance to the economy as a whole, the disconnect between “hard” government numbers and popular perception has spurred me to question that faith. I’ve had the benefit of living in two realms that seem rarely to intersect — one as a Washington insider, the other as an adviser to lenders and investors across the country. Toggling between the two has led me to be increasingly skeptical that the government’s measurements properly capture the realities defining unemployment, wage growth and the strength of the economy as a whole.

…Our research revealed that the data collected by the various agencies is largely accurate. Moreover, the people staffing those agencies are talented and well-intentioned. But the filters used to compute the headline statistics are flawed. As a result, they paint a much rosier picture of reality than bears out on the ground.

Take, as a particularly egregious example, what is perhaps the most widely reported economic indicator: unemployment. Known to experts as the U-3, the number misleads in several ways. First, it counts as employed the millions of people who are unwillingly under-employed — that is, people who, for example, work only a few hours each week while searching for a full-time job. Second, it does not take into account many Americans who have been so discouraged that they are no longer trying to get a job. Finally, the prevailing statistic does not account for the meagreness of any individual’s income. Thus you could be homeless on the streets, making an intermittent income and functionally incapable of keeping your family fed, and the government would still count you as “employed.”

«

This is a remarkable piece, which also shows that “inflation” and “wages” are recorded correctly, but to people who are living in the bottom half of society, things are much worse than the numbers suggest.
unique link to this extract


Learning from my mistakes: The Sun’s 2013 paywall • LinkedIn

Chris Dundan was there when The Sun’s first attempt at a paywall started:

»

Paywalls need exclusivity to function. We worried a lot about locking down the football clips from piracy, for example. The Premier League had a lot of experience in fighting that, but it was a never ending battle and most British pubs contained at least one fan who knew how to illegally stream and questioned why their mates would pay to get clips.

Closer to home it was obvious very early on that The Mirror, The Star and Mailonline were competitors for content and free, and exclusives were very short lived. At one point we stood in the newsroom with a stopwatch and counted eight minutes between The Sun breaking a big scoop and the Mailonline publishing a “fair use” version of the story which sat at the top of the search rankings.

The easiest content to protect was the star columnists who had personal brands and long form writing that was harder to replicate. The best part of the offering that had demonstrable value were the subscriber rewards with discounts on holidays and at retail. When we closed the service it was the rewards club that members asked to continue paying for, which tells you a lot.

Acquisition was pretty strong throughout, because The Sun had a great strength in promotional marketing and the team had learnt a lot about subscription sales. We acquired well over a million customers, but rarely got the subscriber base over 200,000. The drop off from free or discounted trial was recognisably steep. The drop off after the first bill landed was much steeper.

The measure of ‘90 day subscribers’ became the critical measure, and it spent most of the two years in A&E on oxygen surrounded by concerned relatives.

The teams working on the project were well funded, well supported and well motivated. They absolutely bust their asses. For a hundred straight weeks in a row we made the machine better, in a thousand different ways. We saw the much loved marginal gains in all of the KPIs on the dashboard, and we left no stone unturned to do it.

In the end though, you can’t optimise your way out of a black hole, the gravity is too heavy. We were marketing a product at a price point that was material to our customers, and giving them content which was largely available from our competitors for free.

«

Fabulous writing. The Sun+ paywall was knocked down after two years. Maybe this time?
unique link to this extract


The Apple TV app is now available on Android • 9to5Mac

Benjamin Mayo:

»

The Apple TV app is finally making its way to Android users. The app is now available in the Google Play Store for Android phones and tablets. This means Android users can sign up to Apple TV+ and MLS Season Pass directly on their phone for the first time.

The app lets users can log in with their existing Apple Account, or create a new account through the app and subscribe using Google Play billing. The app is rolling out now, supporting Android 10 or newer devices.

The app supports the fundamental Apple TV app features users will be familiar with from the app on Apple devices, including the Continue Watching queue, offline downloads and search. Playback progress syncs across all your devices, so you can start watching a show on your TV and then continue in bed on your Android phone, for instance.

However, the app is limited to Apple TV+, MLS Season Pass and MLB Friday Night Baseball content. iTunes Store purchases and rentals are not shown in the app at all, nor can users cannot access their previously purchased library through the Android app.

In some ways, though, this makes for a cleaner experience with a simple layout of four tabs on the phone app: Apple TV+, MLS, Downloads and Search. On tablet form factors, this interface is presented in a floating sidebar layout.
Rather than a direct port of the iOS app, Apple is using native Android UI components where applicable, such as context menus when long-pressing on an item.

«

Apple TV+ launched in November 2019. Got to the Google Chromecast in February 2021. It’s built into multiple TVs. But there still isn’t a Windows app. You could argue that Apple is just trying to make sure it has plenty of content. (Or perhaps it doesn’t trust Windows users not to pirate everything. Though I suspect it’s all available on torrents anyway.)
unique link to this extract


Trump just gutted the most important number you’ve never heard of • The Washington Post

Cass Sunstein:

»

With the deluge of executive orders in the initial weeks of the second Trump administration, an important directive flew under the radar. It requires the federal government to consider abandoning “the social cost of carbon,” potentially undercutting all climate policymaking.

That is a technical way of signaling something simple and false: Climate change is not real. If the social cost of carbon is treated as zero, then greenhouse gas emissions inflict no damage. Regulations that reduce those emissions have no benefits, which suggests that those regulations should be eliminated.

The social cost of carbon has often been described as the most important number you’ve never heard of. The metric is meant to capture the harm caused by a ton of carbon emissions, making it a foundation of national climate change policy. A lower value would justify weaker regulations, while a higher one would warrant more aggressive policies.

Under the Obama administration, in which I served, the social cost of carbon was relatively modest: around $50. As the Government Accountability Office found, the interagency process that led to that valuation was emphatically apolitical. The calculation, whose use was upheld in federal court, helped support numerous regulations involving fuel economy, energy efficiency and power plants.

To its credit, the first Trump administration maintained a social cost of carbon. But it made significant changes. By far the most important was to adjust the metric so that it would include only domestic damage.

Naturally, the harm inflicted within the United States is a mere fraction of that imposed on the world. The $50 figure immediately fell to about $7, meaning that the benefits of emissions reductions would be a lot smaller.

«

The GOP really has been looking around at absolutely everything it hates. And yet, climate change doesn’t care if we respond to it or not. But we might.
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

2 thoughts on “Start Up No.2383: the talkative AI future, Johansson demands deepfake law, AI summaries no good for news, and more

  1. Sadly, I think most of those people who ignore reality will be long gone before it hits them. It seems that the last years have not been enough and it has to get a LOT more serious.

Leave a reply to Axel Cancel reply

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