A software update intends to make the Autopilot in two million Tesla cars in the US safer – but can it improve the drivers? CC-licensed photo by pedrik 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. Look, no hands! I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.
ChatGPT to summarize Politico and Business Insider articles in ‘first of its kind’ deal • The Guardian
Blake Montgomery (and agencies):
»
Axel Springer, the publisher of Business Insider and Politico, said on Wednesday it was partnering with OpenAI, which will pay the German media group to allow ChatGPT to summarize current articles in responses generated by the chatbot.
“ChatGPT users around the world will receive summaries of selected global news content from Axel Springer’s media brands,” which also includes the German tabloid Bild, the two companies said in a statement.
The chatbot’s answers will include material otherwise kept behind a paywall and offer “links to the full articles for transparency and further information”, they said. Axel Springer will be paid for making its content available to the US artificial intelligence firm, a spokesman for the media group told AFP. The deal is valid for several years and does not commit either side to exclusivity, leaving them free to sign new agreements, the spokesman said without giving more detail.
…The Axel Springer spokesman said that with the advent of platforms such as Google and Facebook, media companies had missed the opportunity to establish a new source of income “and we are all still running after the money”.
The partnership with OpenAI was “the first of its kind”, Axel Springer’s CEO, Mathias Döpfner, said in the statement.
“We want to explore the opportunities of AI empowered journalism – to bring quality, societal relevance and the business model of journalism to the next level,” Döpfner said.
«
Enterprising move by Springer: if it is getting paid by OpenAI and will continue to get paid, it might have made a smart move.
unique link to this extract
The cable news kayfabe is dead • Nieman Journalism Lab
Ben Collins:
»
Too many news institutions have been sucked into the theatre of the absurd, and people are looking for champions who allude to that.
The good news is, the kids see the kayfabe of it all. They are now aware of the game within the game.
They aren’t watching the news because of it, but they are interested in how that kayfabe frames the ever-increasing powerlessness they feel in the ambient horrors playing in the background of their daily lives.
The cable news kayfabe, as you know, goes as follows: a powerful person says something shocking — a far-right politician launches a nativist talking point, or a billionaire speaks of a threat of financial doom if demands aren’t met. A more reasonable voice reacts, frequently asking to please remove the racism, but conceding an underlying but unprovable point that should never have been conceded. The story is framed as reaction to the initial statement, no matter how ludicrous or even impossible that initial statement is.
The reality, in this situation, is and has never been a consideration. This is how you build a world of kayfabe, and you have to turn off parts of your brain to enter it.
You can apply this to countless stories that required a nuance many mainstream news outlets have so far refused to deploy: the apocalyptic hype cycle and predictable burnout of AI, the Israel-Hamas war, TikTok’s influence on American culture and politics.
There are strict parameters on how we talk about each of these things. It’s a sort of news kayfabe: a binary, good guy–bad guy game we’ve invented that has all too often been infiltrated by all bad guys.
…Here’s the good news: The faster you stop playing those games, the faster you stop making Faustian bargains for access, the faster you stop presenting weird false choices that leave out seemingly every American under 40 years old, the faster we can gain their trust back.
The other good news is that they’re right. The kind of news that they want does not have these internecine power structures and middle school-style influence games built into it. That’s not a lot to ask.
«
Certainly seems to be true of the American media ecosystem Collins lives in; I’d like to think that other countries do less badly. But he makes many good points.
unique link to this extract
Dear journalists: here’s how to talk to women on the Hugging Face team • LinkedIn
Emily Witko (and coworkers):
»
We here at Hugging Face have been noticing a concerning trend in tech journalism. The real achievements of women on our team often get overshadowed by a focus on personal, and sometimes very intrusive, details that aren’t relevant to their work. It’s time for that to change. Here’s a set of guidelines that the team has put together, aiming for a more respectful and balanced approach to reporting:
Highlight achievements: Center your articles on professional accomplishments, not on personal attributes like looks, age, or family status. This one is pretty self-explanatory, right folks?
Avoid gendered language: At the moment, we see lots of over-associating women with certain words and concepts, such as ‘children’ and ‘family.’ Proofread your articles to eliminate gendered descriptions that may unintentionally reinforce stereotypes.
• Problematic: Jane Janey, despite being a mother of two, has surprisingly managed to lead her team to develop a groundbreaking AI algorithm.
• Good: Jane Janey, an accomplished leader in her field, has successfully spearheaded the development of a groundbreaking AI algorithm with her team.Respect privacy: Honour the interviewee’s wishes regarding the disclosure of personal information.
«
There’s quite a few more advisements. I’m slightly surprised that journalists – and especially American journalists – would need telling any of this.
unique link to this extract
Russian foreign intelligence service spotted exploiting JetBrains vulnerability • The Record
Jonathan Greig:
»
Government agencies in the US, Poland and the UK said on Wednesday that Russia’s Foreign Intelligence Service (SVR) has been exploiting a vulnerability that was exposed earlier this year in a popular product from Czech software giant JetBrains.
Officials said they have notified dozens of companies across the US, Europe, Asia and Australia after discovering hundreds of compromised devices.
The agencies attributed the attacks to hackers within the SVR known as APT29 — also tracked by cybersecurity researchers as CozyBear or Midnight Blizzard — and said the “large scale” campaign began in September.
Microsoft previously said North Korean hackers were exploiting the bug — labeled CVE-2023-42793 — in September. It affects a product called TeamCity, which is used by developers to test and exchange software code before its release.
Now the SVR has been spotted “using the initial access gleaned by exploiting the TeamCity CVE to escalate its privileges, move laterally, deploy additional backdoors, and take other steps to ensure persistent and long-term access to the compromised network environments,” Wednesday’s alert said.
“Generally, the victim types do not fit into any sort of pattern or trend, aside from having an unpatched, Internet-reachable JetBrains TeamCity server, leading to the assessment that SVR’s exploitation of these victims’ networks was opportunistic in nature and not necessarily a targeted attack.”
The organizations attacked include an energy trade association; companies that provide software for billing, medical devices, customer care, employee monitoring, financial management, marketing, sales, and video games; as well as web hosting companies, tool manufacturers, and small and large IT companies.
«
A story which comes the same day that the British government was warned that the UK is very vulnerable to ransomware attacks. It doesn’t rain but it pours.
unique link to this extract
Tesla recalls two million cars with ‘insufficient’ Autopilot safety controls • The Washington Post
Leo Sands, Aaron Gregg and Faiz Siddiqui:
»
Tesla is recalling more than two million vehicles to fix Autopilot systems that US safety regulators determined did not have enough controls to prevent misuse, the largest recall of Tesla’s driver-assistance software to date.
The National Highway Traffic Safety Administration said Tesla’s method of ensuring drivers are still paying attention while the driver-assistance system is activated is “insufficient.”
“There may be an increased risk of a crash,” the agency wrote, in some situations when the system is engaged “and the driver does not maintain responsibility for vehicle operation and is unprepared to intervene as necessary or fails to recognize when Autosteer is canceled or not engaged.”
The recall comes days after The Washington Post published an investigation that found Teslas in Autopilot had repeatedly been involved in deadly crashes on roads where the software was not intended to be used.
NHTSA said Tesla will send out a software update to fix the problems affecting its 2012-2023 Model S, 2016-2023 Model X, 2017-2023 Model 3, and 2020-2023 Model Y vehicles, effectively encompassing all Tesla vehicles equipped with Autopilot on US roads. Autopilot is a standard feature on Tesla’s vehicles; only some early Tesla models are not equipped with the software.
…The software update, which was to be deployed on “certain affected vehicles” starting Dec. 12, will add extra controls and alerts to “encourage the driver to adhere to their continuous driving responsibility,” the recall report said. The update also will include controls that prevent Autosteer from engaging outside of areas where it is supposed to work as well as a feature that can suspend a driver’s Autosteer privileges if the person repeatedly fails to stay engaged at the wheel.
«
“Recall” in this context doesn’t actually mean “take back into its factories” – at least for Tesla, which can (as the story says) just send out a software update over the air; nobody’s car has to move an inch. The problem of inattentive or unready drivers, though, can’t be so easily fixed.
unique link to this extract
20 things we learned from the Epic v. Google trial • The Verge
Sean Hollister:
»
I have spent 15 days reporting live from the Epic v. Google trial: an antitrust dispute over whether Google’s Android app store is an unfair monopoly. I’ve watched a parade of witnesses go by, including Epic CEO Tim Sweeney and Google CEO Sundar Pichai. We’re now in a weeklong break before both parties return on December 11th to make their closing arguments, after which a jury will decide who’s right. I’ve chronicled every major thrust, parry, and riposte leading up to that in our Verge StoryStream, writing nearly 600 dispatches from the courtroom so far.
But who’s got the time to dig through all that, am I right?
So here are straightforward versions of the 20 most interesting things we’ve learned — starting with the fact that Epic could win the whole thing.
«
This was last week, but Hollister was right about Epic winning, and a number of the other details are just as interesting – notably No.9: “This trial destroyed any notion that Google treats developers fairly and equally”. Also No.13: profit margins for the Play Store were north of 70%.
unique link to this extract
Cop28 is a farce rigged to fail, but there are other ways we can try to save the planet • The Guardian
George Monbiot:
»
Since this horrible farce [of Cop] began 31 years ago, plenty of people have proposed reforms. The proposals fall into three categories. One is to improve the way consensus decisions are made. Well-meaning as these are, they’re futile: you can tweak the process, but it will remain dysfunctional.
Another approach is to replace consensus decision-making with voting, an option that remains, in draft form, in the UN rules. The obvious objection is that a majority would impose decisions on other nations. But this reflects a narrow conception of what voting could do. There are plenty of ways of ensuring everyone can be heard, without relying on crude binary choices. One of the most promising is the Borda count, a decision-making method first proposed in 1435.
The modified Borda count developed by the de Borda Institute looks especially useful. First, the delegates agree on what the principal issues are. These are then turned into a list of options, on which everyone is asked to agree (the options could range from the immediate phase-out of fossil fuels to planetary Armageddon). The options are listed on a ballot paper, and each delegate is asked to rank them in order of preference. A scoring system awards points for every ranking. The more options a delegate ranks, the more points each one is worth to them. This enables complex decisions to be made without excluding anyone.
The third approach, which could run alongside the second, is to bypass the Cop process by developing new binding treaties. The professor of environmental politics Anthony Burke suggests an approach modelled on the 2017 treaty on the prohibition of nuclear weapons, the 1997 anti-personnel mine ban convention and the 2008 convention on cluster munitions. In these cases, states and citizens’ groups frustrated with a lack of progress began building treaties without the participation of the powerful nations – the US in particular – that sought to resist them. They developed enough momentum not only to push the treaties through the UN general assembly, but also to establish new diplomatic norms that made defiance of the treaties much harder to justify, even for nations that refuse to ratify them.
«
In retrospect, it’s absolutely incredible that the Montreal Protocol – to drastically cut CFC production and use – was agreed and implemented. Could it be done today?
unique link to this extract
Bluesky isn’t a mere Twitter clone, says CEO Jay Graber • Fast Company
Harry McCracken:
»
As 2023 winds down, Bluesky, which still hasn’t opened up to all comers, has grown to 2.3 million users. Like Mastodon, Threads, and other refuges for Twitter expatriates, it hasn’t become the sort of one-stop conversation megahub that Twitter once was. Instead, an eclectic subset of the Twitter masses has landed there. CEO Jay Graber says the 30-person Bluesky team tried to recreate “the platonic ideal of microblogging as it once was” and calls out some constituencies who have bonded with the service: meme lovers, writers and artists, and people who find enforced pithiness to be a fun, creative challenge. (Bluesky has a 300-character limit.)
Why has Bluesky kept its invite system in place even though that means many would-be members still haven’t gotten in? “To be honest, we didn’t have the capacity to absorb all of that in a day,” says Graber. “It’s a full-time effort for engineers to scale up the service at the level that we were going. We actually have not had significant downtime, which is a testament to this controlled-growth strategy—you know, Twitter early on had a lot of failwhales.”
Whether you’re already on Bluesky or not, thinking of it purely as a Twitter-esque app misses the point. The app is just a testbed for the protocol the startup has been building to help nudge us out of the era of centralized social networking. And Graber—who got her job as CEO in 2021 after DMing Dorsey to express enthusiasm for the Bluesky project—says that the founding goal is soon to get its first major real-world test.
«
That test will be in early 2024, when decentralisation will be tried, and the invite system will become redundant.
unique link to this extract
Apple is holding the final nail for X’s coffin • Bloomberg
Dave Lee:
»
To celebrate the return of such a man [as Infowars’ Alex Jones], Musk joined Jones in a live audio chat on Sunday. Other attendees included Andrew Tate, who faces rape and sex-trafficking charges, and Republican presidential hopeful Vivek Ramaswamy, who last week rattled off a raft of unfounded conspiracy theories on the debate stage. With saner minds abandoning X, this is the clientele that’s left.
Apple’s marketing team clearly wants no part of it. It’s among the companies that have decided to “pause” advertising on the platform, having previously been its biggest spender. While not mentioned directly, Apple was implied among the group — along with Disney, Walmart and others — that Musk recently told to go f— themselves.
But Apple holds an even greater power than that. Having decided X is too dangerous for its brand, the reinstatement of Jones now forces it to confront a bigger question: When does X become too dangerous for Apple’s users? Or when do Apple’s supposed corporate principles demand it no longer take the 30% it enjoys from sign-ups to X’s premium services? (The company did not respond to a request for comment.)
Apple has answered this question before. Jones’s InfoWars app was banned from the App Store in 2018. His content, the company said at the time, fell afoul of “defamatory, discriminatory, or mean-spirited content, including references or commentary about religion, race, sexual orientation, gender, national/ethnic origin, or other targeted groups, particularly if the app is likely to humiliate, intimidate, or place a targeted individual or group in harm’s way.” The company did not specify which specific material had forced it to act. The ban appears to remain in place today.
…The fallout would be several magnitudes greater were Apple to take any action against X. But on its current trajectory — with Musk himself amplifying some of the platform’s most vicious elements — there will come a time when inaction would be just as damaging to Apple.
«
Musk’s X 2023 ad sales projected to slump to about $2.5bn • Bloomberg via MSN
Kurt Wagner:
»
Elon Musk’s X, the social network formerly known as Twitter, is on track to bring in roughly $2.5bn in advertising revenue in 2023 — a significant slump from prior years, according to people familiar with the matter.
X generated a little more than $600m in advertising revenue in each of the first three quarters of the year, and is anticipating a similar performance in the current period, according to a person familiar with the numbers. That compares to more than $1bn per quarter in 2022.
Ad sales currently make up between 70% and 75% of X’s total revenue, according to two people with knowledge of the matter. That would imply 2023 sales of roughly $3.4bn, including sales from subscriptions and data licensing deals.
The previously unreported sales figures underscore with greater clarity advertisers’ unease with how X is handling content moderation under Musk, and in particular the new owner’s posts that amplify antisemitic and other extremist views.
X executives had originally targeted $3bn in revenue from advertising and subscriptions in 2023, but will fall far short of that number, one of the people said. The annual number is still in flux because the holiday quarter isn’t yet over.
“This presents an incomplete view of our entire business, as the sources you’re relying on for information are not providing accurate and comprehensive details,” said Joe Benarroch, head of business operations for X.
«
For comparison, revenue in the year before Musk was $5bn (though it wasn’t profitable then either). No publisher would think they were making a roaring success of things if they halved income, though he may have halved – or cut even further – the outgoings.
What’s notable is that Wagner actually got a response from a human at the company. That’s how significant it sees both Bloomberg, and this revelation. Anyway, perhaps he can blame it all on whoever let that sink in, and the bozo who was holding it.
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
The Overspill: when there's more that I want to say 