The departures from the executive ranks at Twitter are continuing, with two key people leaving at the end of last week. CC-licensed photo by CeltikipoohCeltikipooh 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 was another post coming last Friday at the Social Warming Substack. Go have a read! Free signup.
A selection of 10 links for you. Use them wisely. I’m @charlesarthur on Twitter. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.
The AI founder taking credit for Stable Diffusion’s success seems to have a history of exaggeration • Forbes
Kenrick Cai and Iain Martin:
»
AI researchers with whom [Stability AI CEO and founder Emad] Mostaque worked told Forbes he claimed credit he did not earn or deserve. And when pressed, Stability spokesperson Motez Bishara admitted to Forbes that Stability had no special deal with Amazon.
Mostaque’s other mischaracterizations to investors include multiple fundraising decks seen by Forbes that presented the OECD, WHO and World Bank as Stability’s partners at the time — which all three organizations deny. Bishara said the company could not comment on the presentations “without knowing the exact version,” but that they were accompanied by additional data and documentation.
Inside the company, wages and payroll taxes have been repeatedly delayed or unpaid, according to eight former employees, and last year the UK tax agency threatened to seize company assets. (“There were several issues that were expeditiously resolved,” Bishara said.)
…In responding to a detailed list of questions, Mostaque shared a statement saying that Stability had not historically prioritized the “systems and processes” underpinning the fast-growing startup. “We recognize our flaws, and we are working to improve and resolve these issues in an effective and compassionate manner,” he wrote.
AI experts and prospective investors have been privately expressing doubts about some of Mostaque’s claims for months now. Despite Silicon Valley’s sudden, insatiable appetite for AI startups, a number of venture capitalists told Forbes that the Stability founder has been struggling to raise hundreds of millions more in cash at a roughly $4bn valuation. Mostaque publicly claimed last October that annualized revenue had surpassed $10m, but insiders say sales have not improved (Bishara said the October number was “a fair assessment of anticipated revenues at the time,” and declined to comment on current revenue). “So many things don’t add up,” said one VC who rejected Mostaque’s funding overtures.
«
I’m wary of quoting the article in full because of British libel law, but it’s really not complimentary to Mostaque. Though one claim – that his wife, who’s head of PR, took thousands of pounds from the company – is completely explicable as repayment of a loan made previously to the company. You can see that when some VC money comes in, you might want to collect on the money put in months or years ago. Loans by directors to and from their companies are completely routine in startups.
unique link to this extract
Air Force official’s story of killer AI was a hypothetical • Business Insider
Charles Davis and Paul Squire:
»
An Air Force colonel who oversees AI testing used what he now says is a hypothetical to describe a military AI going rogue and killing its human operator in a simulation in a presentation at a professional conference.
But after reports of the talk emerged Thursday, the colonel said that he misspoke and that the “simulation” he described was a “thought experiment” that never happened.
Speaking at a conference last week in London, Col. Tucker “Cinco” Hamilton, head of the US Air Force’s AI Test and Operations, warned that AI-enabled technology can behave in unpredictable and dangerous ways, according to a summary posted by the Royal Aeronautical Society, which hosted the summit.
As an example, he described a simulation where an AI-enabled drone would be programmed to identify an enemy’s surface-to-air missiles (SAM). A human was then supposed to sign off on any strikes.
The problem, according to Hamilton, is that the AI would do its own thing — blow up stuff — rather than listen to its operator.
“The system started realizing that while they did identify the threat,” Hamilton said at the May 24 event, “at times the human operator would tell it not to kill that threat, but it got its points by killing that threat. So what did it do? It killed the operator. It killed the operator because that person was keeping it from accomplishing its objective.”
But in an update from the Royal Aeronautical Society on Friday, Hamilton admitted he “misspoke” during his presentation. Hamilton said the story of a rogue AI was a “thought experiment” that came from outside the military, and not based on any actual testing.
«
I read the blogpost, like all the other people who reported on this, and at no point did it suggest that this was a simulation in his mind. But it did echo some of the examples recorded in “The surprising creativity of digital evolution: a collection of anecdotes from the evolutionary computation and artificial life research communities“, a PDF full of weird stories. (I referenced it in Social Warming. It’s a fascinating read.)
Of course the hype around this story echoed the concerns about AI – extinction! – that have been going round. So no surprise that nobody called Hamilton.
unique link to this extract
AI entertainment made to order: you think you want that? • At the Mountains of Sadness
Mike Drucker is a screenwriter:
»
On its face, entertainment made to order sounds nice, like finding porn of your favorite proper noun. In theory, it means that everything you watch will be specifically created for you and will cater to your tastes. You won’t have to scroll through a streaming service, quit the app, and then scroll another streaming service. You could say, “A romantic comedy starring Arianna Grande and Greta Garbo” and a machine would spit it out for you. And it’ll be perfect.
A couple things though.
First – and most obvious – none of us know what the fuck we want. I don’t mean that in some vast “We, the gatekeepers know what you want more than you do” way. I mean it in the literal sense of none of us know what the fuck we actually want from moment to moment. We aren’t scrolling endlessly through streaming apps because there’s nothing that appeals to us. We’re scrolling endlessly through streaming apps because we aren’t sure what we want to watch at that specific moment. There’s thousands of things that appeal to us we just skim right by.
You do it on Spotify all the time when a playlist you made for yourself hits a song you absolutely do not want to hear.
Like, I’m a fan of Law & Order, but I don’t want to watch it all the time. Sometimes I’ll skip right past it. No Law & Order for me! No thank you! I’m all set on what happens in a version of the Manhattan legal system featuring courtrooms with natural sunlight. Except, sometimes I think I want to watch Law & Order. And I really do and love it! Other times I don’t. And then I turn it off. And then I look for something else, even though my brain was certain I wanted Law & Order.
Rather, it’s what we didn’t think we’d like that surprises us. Shows we think, “What the hell” and then end up loving. Documentaries that look boring as dog shit turn into thrilling mysteries. You finally got around to watching that movie everyone won’t shut up about with that actor you hate and – guess what? – you actually loved it. You didn’t know you would. But you did.
«
It’s an excellent point. My expectation is that the AI systems will just churn out unlimited stuff and it’ll just be semi-random what you pick; serendipity might be the next phase of the internet.
unique link to this extract
The world’s peak population may be smaller than expected • The Economist
»
few have noticed a wealth of new data that suggest that Africa’s birth rate is falling far more quickly than expected. Though plenty of growth is still baked in, this could have a huge impact on Africa’s total population by 2100. It could also provide a big boost to the continent’s economic development. “We have been underestimating what is happening in terms of fertility change in Africa,” says Jose Rimon II of Johns Hopkins University. “Africa will probably undergo the same kind of rapid changes as east Asia did.”
The UN’s population projections are widely seen as the most authoritative. Its latest report, published last year, contained considerably lower estimates for sub-Saharan Africa than those of a decade ago. For Nigeria, which has Africa’s biggest population numbering about 213m people, the UN has reduced its forecast for 2060 by more than 100m people (down to around 429m). By 2100 it expects the country to have about 550m people, more than 350m fewer than it reckoned a decade ago.
Yet even the UN’s latest projections may not be keeping pace with the rapid decline in fertility rates (the average number of children that women are expected to have) that some striking recent studies show. Most remarkable is Nigeria, where a UN-backed survey in 2021 found the fertility rate had fallen to 4.6 from 5.8 just five years earlier. This figure seems to be broadly confirmed by another survey, this time backed by USAID, America’s aid agency, which found a fertility rate of 4.8 in 2021, down from 6.1 in 2010. “Something is happening,” muses Argentina Matavel of the UN Population Fund.
If these findings are correct they would suggest that birth rates are falling at a similar pace to those in some parts of Asia, when that region saw its own population growth rates slow sharply in a process often known as a demographic transition.
«
It’s not just a fiscal fiasco: greying economies also innovate less • The Economist
»
it is on productivity that demographic decline may have the most troubling effect. Younger people have more of what psychologists call “fluid intelligence”, meaning the ability to solve new problems and engage with new ideas. Older people have more “crystallised intelligence”—a stock of knowledge about how things work built up over time. There are no precise cut-offs, but most studies suggest that fluid intelligence tends to peak in early adulthood and to begin to decline in people’s 30s. Both types of intelligence are useful: companies, industries and economies need both youngsters able to respond to new challenges and seasoned veterans with a detailed understanding of their trade. But the two are not of equivalent value when it comes to innovation.
In research published in 2021, Mary Kaltenberg and Adam Jaffe, both economists, and Margie Lachman, a psychologist, used a database of 3m patents filed over more than 40 years to explore the relationship between innovation and age. Depending on the scientific discipline, the authors note that patenting rates peak in a researcher’s late 30s and early 40s. The rates of patenting then decline only gradually through their 40s and 50s.
But for disruptive innovations, which fundamentally change a scientific field, the picture is very different. The researchers used a measure of disruptiveness based on the number of citations of a given invention in future patents. If a particular patent is cited by subsequent inventors, but that patent’s technological predecessors are not, it is categorised as a disruptive rather than an incremental innovation. For instance, the work of Kary Mullis, a Nobel-prize-winning biochemist, on polymerase chain reactions underpins much modern genetic and medical testing. After Mullis published his work, citations of prior technologies in the same field nosedived. Mr Jaffe, Ms Kaltenberg and Ms Lachman find that patents filed by the very youngest inventors are much more likely to be completely novel, discipline-changing innovations, and that as inventors age the patents they file become increasingly incremental.
«
I think these articles were written by Mike Bird, but The Economist doesn’t do bylines, so it’s just surmise based on his tweets.
unique link to this extract
The big question of how small chips can get • Financial Times
Tim Bradshaw and Anna Gross:
»
chipmakers now face a daunting challenge. Moore’s prediction has fallen behind schedule and the cadence is now closer to three years. The latest 3-nanometre chips being mass produced for this year’s iPhones will be followed by what some see as an even bigger leap forward to 2nm by 2025. “But once you get to 1.5nm, maybe 1nm, Moore’s Law is 100% dead,” says Ben Bajarin, a technology analyst at Silicon Valley-based Creative Strategies. “There’s just no way.”
Chip engineers have defied forecasts of an end to Moore’s Law for years. But the number of transistors that can be packed on to a silicon die is starting to run into the fundamental limits of physics. Some fear manufacturing defects are rising as a result; development costs already have. “The economics of the law are gone,” says Bajarin.
That has sent chip designers into a scramble over the past few years for alternative ways to sustain advances in processing power, ranging from new design techniques and materials to using the very AI enabled by the latest chips to help design new ones.
…one consequence of chip innovation being more narrowly focused is that any breakthroughs tend to be more zealously guarded and less transferable to the wider market.
“Through the 1990s and early 2000s, cost per transistor and the ability to build more complex chips was roughly free to the entire industry,” says [leader of the US National Network for Critical Technology Assessment, Hassan] Khan. “[Now] computation is less of a general purpose technology . . . If I’m optimising chips for AI, that might make GPT more efficient or powerful but it may not spill over into the rest of the economy.”
Another key delta of innovation is in chip “packaging”. Instead of printing every component on to the same piece of silicon, to create what’s known as a “system on a chip”, semiconductor companies are now talking up the potential for “chiplets” that allow smaller “building blocks” to be mixed and matched, opening up new flexibility in design and component sourcing.
«
Could Ozempic also be an anti-addiction drug? • The Atlantic
Sarah Zhang:
»
As semaglutide [aka Ozempic] has skyrocketed in popularity, patients have been sharing curious effects that go beyond just appetite suppression. They have reported losing interest in a whole range of addictive and compulsive behaviors: drinking, smoking, shopping, biting nails, picking at skin. Not everyone on the drug experiences these positive effects, to be clear, but enough that addiction researchers are paying attention. And the spate of anecdotes might really be onto something. For years now, scientists have been testing whether drugs similar to semaglutide can curb the use of alcohol, cocaine, nicotine, and opioids in lab animals—to promising results.
Semaglutide and its chemical relatives seem to work, at least in animals, against an unusually broad array of addictive drugs, says Christian Hendershot, a psychiatrist at the University of North Carolina at Chapel Hill School of Medicine. Treatments available today tend to be specific: methadone for opioids, bupropion for smoking. But semaglutide could one day be more widely useful, as this class of drug may alter the brain’s fundamental reward circuitry. The science is still far from settled, though researchers are keen to find out more. At UNC, in fact, Hendershot is now running clinical trials to see whether semaglutide can help people quit drinking alcohol and smoking. This drug that so powerfully suppresses the desire to eat could end up suppressing the desire for a whole lot more.
The history of semaglutide is one of welcome surprises. Originally developed for diabetes, semaglutide prompts the pancreas to release insulin by mimicking a hormone called GLP-1, or glucagon-like peptide 1. First-generation GLP-1 analogs—exenatide and liraglutide—have been on the market to treat diabetes for more than a decade. And almost immediately, doctors noticed that patients on these drugs also lost weight, an unintended but usually not unwelcome side effect.
«
That’s going to be an interesting side effect, if confirmed and controllable. Weight loss and addiction reduction? “Wonder drug” gets overused, but this is pretty dramatic.
unique link to this extract
Why trends have lost all meaning • Fast Company
Matt Klein:
»
Trends once meant meaningful social change: an emerging and defining collective thought, behavior, value, or attitude. A shift in society. But today, “mermaidcore” is being named a top trend for this summer.
Trends lost their meaning.
When brands joined the conversations on social media en masse, it set off a chain reaction that would ultimately lead us to devalue the rigorous practice of trend forecasting and the original definition of a “trend.”
As soon as brands created their own accounts the thought became: “If we participate in these discussions, we win culture . . . and then sales.”
It’s unclear if this notion has even been effectively measured or supported, but it often remains the collective hypothesis.
As a result, brands started obsessing over the “trending” story of the day, hashtag, meme of the moment, or core aesthetic. Watching everyone engage in public conversations across social media, many brands believe emulating our “friends” would unlock cultural resonance.
In the process of chasing cool, brands have lost the purpose of analyzing culture. Most of the signals considered “trends” today are really nothing more than frivolous entertainment.
«
I always thought that trends were meant to indicate deep cultural changes that brands and companies would want to engage with. Such as “Tribal Rites of the New Saturday Night“.
unique link to this extract
Twitter’s head of brand safety and ad quality to leave • Reuters
Tiyashi Datta and Sheila Dang:
»
Twitter’s head of brand safety and ad quality, A.J. Brown, has decided to leave the company, according to a source familiar with the matter on Friday, the second safety leader to depart in a matter of days.
The latest departure adds to a growing challenge for new Twitter CEO Linda Yaccarino, even before she steps into the role.
On Thursday, Ella Irwin told Reuters that she resigned from her role as vice president of product for trust and safety at the social media company, where she oversaw content moderation efforts and often responded to users with questions about suspended accounts.
Brown worked on efforts to prevent ads from appearing next to unsuitable content.
Platformer and the Wall Street Journal earlier reported Brown’s departure.
Since Tesla CEO Elon Musk acquired Twitter in October, the platform has struggled to retain advertisers, who were wary about the placement of their ads after the company laid off thousands of employees.
«
The departure of Irwin seems to have been over a film that Musk wanted shown titled “What is a woman?” What’s strange is that Irwin chose to resign over that, rather than all the other things Musk had done. Also, I haven’t heard of any hirings at Twitter; only departures.
unique link to this extract
The Google Pixel Watch is randomly popping open for some • Android Police
Rajesh Pandey:
»
Over on Reddit, reports from Pixel Watch owners about its backplate suddenly falling out have been popping up for at least the last few months. The issue does not seem widespread, but it is also not small enough to be ignored. In most cases, users report their Pixel Watch back came off while removing the wearable from the charging puck. This indicates an issue with the adhesive that holds the back and the watch together.
Contacting Google support has worked for most users, as the company sent a replacement unit to most. However, in some cases, owners had to escalate the issue as support asked for $300 to file a warranty claim or denied a replacement.
This is not the first instance of Google’s devices facing random hardware or build quality issues. There have been complaints from Pixel 7 users about its rear camera glass spontaneously shattering without any apparent reason. With the 2021 Pixel 6 series, there were complaints about its screen randomly cracking.
Google has not yet acknowledged the problem. But with such reports appearing frequently, it’s clear that there’s an issue with the adhesive that glues the smartwatch’s back to the chassis.
«
“Not widespread, but not small enough to be ignored” puts it well. The Pixel Watch will have sold in really quite small volumes, so these reports definitely suggest a design problem.
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 