Start Up No.2004: musicians embrace ChatGPT, wind trumps gas in UK, Twitter’s mystery CEO, fake science proliferates, and more

There’s more bad news for Peloton, which is recalling 2.2m of its exercise bikes in the US over a seat fault. (But not outside the US?) CC-licensed photo by Dana L. Brown 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. What’s missing? I’m @charlesarthur on Twitter. On Mastodon: Observations and links welcome.

Musicians are already using AI more often than we think • Pitchfork

Marc Hogan:


Shawn Everett, the Grammy-winning engineer and producer behind albums by Kacey Musgraves, the War on Drugs, and Alvvays, compares the advent of AI in music to the advent of the electric guitar or sampling. “As far as songwriting and production goes, we’re on the cusp of a wave of something that I don’t think we’ve really seen, maybe ever,” he says.

Everett paid attention in 2020 when OpenAI put out a tool for creating songs in various artists’ styles, complete with vocals. He even experimented with that tool while working on a song by the Killers that has never been released. Everett recalls inputting a chord progression that frontman Brandon Flowers had written and instructing the AI to continue it in the style of Pink Floyd, with a certain emotional tenor, only to have the AI spit out unexpected melodies. “What was happening was so different, and was landing in locations that no human being would normally think of, but it still felt rooted in something familiar,” he says. “I thought it was such a cool song.”

What’s coming next, Everett predicts, will be AI tools that can quickly combine ideas for melody, chords, and rhythm, similar to how programs like Midjourney and Dall-E, which generate images from natural language prompts, have shaken up visual art. Within a year or two, he speculates, the thousands of plugins in digital audio workstations like Pro Tools could merge into a single plugin that seamlessly carries out the user’s verbal requests. As an engineer, he wonders if he will ask the AI to set the EQ for a particular drum style—say, Metallica’s—or if the tech will eventually be able to spit out a replicated Lars Ulrich drum performance that sounds better than any drums he (or anyone else, for that matter) could have mic’d. “Obviously that’s a horrifying scenario for a lot of people, but it’s probably gonna happen,” Everett says. 


Musicians can be the quickest to embrace new technology, and it definitely sounds like this has tickled their fancy.
unique link to this extract

Wind is main source of UK electricity for first time • BBC News

Esme Stallard:


Wind turbines have generated more electricity than gas for the first time in the UK.

In the first three months of this year a third of the country’s electricity came from wind farms, research from Imperial College London has shown. National Grid has also confirmed that April saw a record period of solar energy generation.

By 2035 the UK aims for all of its electricity to have net zero emissions.

“There are still many hurdles to reaching a completely fossil fuel-free grid, but wind out-supplying gas for the first time is a genuine milestone event,” said Iain Staffell, energy researcher at Imperial College and lead author of the report.

The majority of the UK’s wind power has come from offshore wind farms. Installing new onshore wind turbines has effectively been banned since 2015 in England.

Under current planning rules, companies can only apply to build onshore wind turbines on land specifically identified for development in the land-use plans drawn up by local councils. Prime Minister Rishi Sunak agreed in December to relax these planning restrictions to speed up development.


Onshore wind isn’t going to happen until we have a different administration, let’s admit.
unique link to this extract

Peloton shares slide after it recalls two million exercise bikes • The New York Times

Lora Kelley:


Peloton, the maker of home exercise equipment, said on Thursday that it was recalling 2.2m exercise bikes, an announcement that sent its stock lower.

The company’s shares tumbled nearly 9% by the market close and have plunged more than 20% this month.

The company had received 35 reports of seat posts breaking and detaching from the original model of its bike during use, according to a recall notice from the Consumer Product Safety Commission.

Peloton is voluntarily recalling Model PL-01 bikes that were sold from January 2018 to May 2023 in the United States, and is offering customers replacements for the bike’s seat posts that can be installed at home, the company said in a statement on its website Thursday morning.

“For Peloton, it was important to proactively engage the C.P.S.C. to address this issue,” the company wrote. “We worked cooperatively with them to identify today’s approved remedy.”

The decision to recall the bikes is a turnabout for Peloton, which in the past has resisted recalling its equipment.


I was wondering: so how many bikes has Peloton sold in the US? And the answer: 2.2m. It’s recalling all of the ones sold in the US. But this, in turn, raises the question of why the ones sold outside the US, which one presumes largely have the same design and parts (especially when it comes to seat posts) aren’t being recalled too. Or is that the next step?

Either way, this could be the killer blow for a company that’s been struggling for a while.

unique link to this extract

Twitter has a new mystery CEO • The Verge

Mitchell Clark, Emma Roth and Jay Peters:


Elon Musk has announced plans for a new Twitter CEO but hasn’t said who it is. In a tweet on Thursday, Musk says that he has “hired a new CEO for X/Twitter” and that “she will be starting in ~6 weeks.” Musk will instead assume the role of executive chair and chief technology officer, “overseeing product, software & sysops” of Twitter.

While Musk may soon no longer be CEO, he still owns the company, which he has renamed “X.” It seems unlikely that giving someone else one specific title will make Twitter any less of a wild ride. Musk became “Chief Twit” last October, when he closed his acquisition of the company, followed by the immediate firing of large portions of its executive staff and thousands of other employees.


Follows the poll back in December asking if he should step down, which got a huge number of people saying damn yes.

The replacement isn’t Sheryl Sandberg (of Facebook), she’s told journalists. Who would be mad enough to take a job where Musk is both your boss (as exec chair) and underling (as CTO)? The news may be out by the time you read this, but wasn’t on Thursday evening. Kara Swisher’s suggestion was Linda Yaccarino of NBC.
unique link to this extract

Elon Musk broke Twitter’s Ratio—the yardstick off all-out ideological insult warfare • Slate

Alex Kirshner:


It might look like a certain kind of victory for Musk that Twitter is now this way: the site is a bit more ideologically attuned to what seems to be his liking, and it’s harder to use the wisdom of the crowd, rather than of those who pay $8 a month, to puncture ridiculous talking points in a visible forum.

But this dynamic ultimately breaks bad for Musk and his buddies, too. It gets harder to warn someone’s followers about the Woke Cancel Culture Mob on the internet, and to position oneself as the last big defender of free speech standing up against the vile left, if a person needs to read through 500 replies to a terrible Ben Shapiro tweet to see someone making a confrontational point in response.

The Twitter experience also just gets a bit less fun, for everyone. Someone on the left who doesn’t pay for Twitter has less incentive to look at tweets they think are terrible, and then craft a reply they think is cutting, if they think there’s a slim chance anyone will read it. And conservatives, who seem more likely to pay for Twitter Blue, will eventually get bored. Twitter’s benefit to them is not that it’s an echo chamber where any right-wing line is met with hugs and kisses.

There are numerous conservative social media websites, and none has gotten anywhere near Twitter’s popularity. There is no Fox News of conservative social media, because much of the fun for right-wing internet users is having an allegedly woke mob to argue with. It may not have happened yet, but it will eventually get stultifying for a huge mass of Twitter Blue subscribers to gather in the replies to talk-show hosts’ tweets to agree that socialism is bad. This kind of poster needs someone to fight with, and by tilting the playing field, Musk has cut down on fighting.

That seems like a good thing for people in two groups: those who would like to spend less time getting their blood pressure up while looking at the internet, and those who would like to see Musk’s Twitter investment degraded until the point of a wipeout.


Some people are in both groups. Just sayin’.
unique link to this extract

Into Thin AirPods • Defector

Casey Johnston lost her AirPods. But not to worry! They were findable with the Find My… app, weren’t they? Well:


when I did open the [Find My] app, there the AirPods were—well, there they weren’t, but also there they were—pinging from a town about 30 minutes away in decent traffic. I zoomed in on the map until it resolved into individual residences. The AirPods appeared to be posted up on a dead end street, squarely in someone’s house. Find My wouldn’t commit to an address, but by cross-referencing Google Maps and a nearby BMW dealership, I was able to triangulate a building number.

The AirPods weren’t in the wind, as lost or stolen objects had been my entire life. They were right there. They were close. They were obtainable. I’d known going into this relationship that I would lose them; until this moment, I hadn’t thought about the possibility that I’d be able to redeem myself by finding them again.

This is the part where I say I’m aware that everyone—Apple, law enforcement, any friends with good judgment within earshot—strenuously discourages ever, under any circumstances, trying to do vigilante justice with the Find My app. If you so much as mention the possibility, like four people will jump out of the woodwork with stories about someone they knew who was shot or assaulted trying to confront a thief in the act. I’d like to emphasize that I’m firmly on the side of reason, and a steadfast believer that having crime done to me is not an occasion to show off how brave I am.

But! I have watched Veronica Mars so many times. I dream idly of mysterious cases falling into my lap, and solving them through the careful piecing together of data, clues, and information, plus the judicious application of wiles and streetwise know-how. And, honestly, I did want my ridiculously expensive AirPods back.


Thus begins a terrific story of pursuit.
unique link to this extract

Fake scientific papers are alarmingly common • Science

Jeffrey Brainard:


When neuropsychologist Bernhard Sabel put his new fake-paper detector to work, he was “shocked” by what it found. After screening some 5000 papers, he estimates up to 34% of neuroscience papers published in 2020 were likely made up or plagiarized; in medicine, the figure was 24%. Both numbers, which he and colleagues report in a medRxiv preprint posted on 8 May, are well above levels they calculated for 2010—and far larger than the 2% baseline estimated in a 2022 publishers’ group report.

“It is just too hard to believe” at first, says Sabel of Otto von Guericke University Magdeburg and editor-in-chief of Restorative Neurology and Neuroscience. It’s as if “somebody tells you 30% of what you eat is toxic.”

His findings underscore what was widely suspected: Journals are awash in a rising tide of scientific manuscripts from paper mills—secretive businesses that allow researchers to pad their publication records by paying for fake papers or undeserved authorship. “Paper mills have made a fortune by basically attacking a system that has had no idea how to cope with this stuff,” says Dorothy Bishop, a University of Oxford psychologist who studies fraudulent publishing practices. A 2 May announcement from the publisher Hindawi underlined the threat: It shut down four of its journals it found were “heavily compromised” by articles from paper mills.

Sabel’s tool relies on just two indicators—authors who use private, noninstitutional email addresses, and those who list an affiliation with a hospital. It isn’t a perfect solution, because of a high false-positive rate. Other developers of fake-paper detectors, who often reveal little about how their tools work, contend with similar issues.

…Sabel’s tool correctly flagged nearly 90% of fraudulent or retracted papers in a test sample. However, it marked up to 44% of genuine papers as fake, so results still need to be confirmed by skilled reviewers.


Those seem like rather broad brush strokes with which to paint things as fake.
unique link to this extract

Finding geolocation leads with Bellingcat’s OpenStreetMap search tool • bellingcat

Logan Williams:


Sometimes, the trickiest part of geolocating a photo can be knowing where to start looking.

In previous Bellingcat investigations, starting points have involved reverse image search, searching Google Earth for soil that is the right colour, looking for minarets and even identification of plants.

Bellingcat has built a new tool for searching OpenStreetMap data to help geolocate images and identify starting points for geolocation investigations, based on objects and structures you can identify in an image. You can think of it as a dramatically simplified version of the Overpass query language tool that some open source researchers may already be familiar with using via Overpass-Turbo. 

Users can sign up for the tool and check it out here.


It’s really clever, using tagged items (“railway” “convenience store” “fountain”) and finding locations where those tags are in close proximity. You’d need to figure out which country your picture is in, and hope that OSM has been updated. (It often has.)
unique link to this extract

AI has potential to be ‘destructive’ to journalism, media tycoon Barry Diller warns • Financial Times

Daniel Thomas, Anna Nicolaou and Laura Pitel:


US media billionaire Barry Diller warned that the use of artificial intelligence would prove “destructive” to journalism unless publishers were able to use copyright law to exert control.

Speaking at the Sir Harry Evans Global Summit in Investigative Journalism in London, Diller said that freely allowing AI access to media content would prove to be a mistake, and that the notion of “fair use” — which can be used to cover copyrighted material in data sets for machine learning — needed to be redefined.

“You can’t have fair use when there is an unfair machine that knows no bounds,” said Diller, who chairs media and internet group IAC.

Media groups have grown concerned about the use of their publications as the basis for creating generative AI. News Corp chief executive Robert Thomson said this year that the group was already seeking financial compensation from an AI company for use of its “proprietary” content.

Diller said on Wednesday that he would work alongside News Corp and German publishing house Axel Springer in trying to protect their journalism from the threat.

“We are leading a group that is going to say we are going to change copyright law if necessary, to work to say that you cannot take our materials or we will litigate. What you publish you have the right to control,” he said.


Yes, driving the price paid for journalism down to the ground is a job for Diller and Murdoch, not the damn machines.
unique link to this extract

Google’s new Magic Editor uses AI to totally transform your photos • The Verge

Jay Peters:


Google’s latest Photos trick is a feature it’s calling Magic Editor, which uses generative AI to let you make major edits to a photo without professional tools. The company revealed the feature at Google I/O 2023.

Google shared a couple examples of Magic Editor in action that are both pretty cool. In one, a photo of a person in front of a waterfall, Google entirely moves the person further to the side of the photo, erases people in the background, and makes the sky a prettier blue. Watch this GIF to see it all happen:

GIF: Google

In another photo, Magic Editor scoots a child on a bench closer to the middle of the photo, which generates “new” parts of the bench and balloons to the left to fill in the space. In this example, Google again makes the sky more vibrant.

It’s impressive stuff — and a logical next step from photo features like Photo Unblur and Magic Eraser. It’s also not quite perfect given leftover artifacts like creases from the bag strap in the waterfall photo and a misplaced shadow under the bench in the second. But perhaps most importantly, it’s just the latest opportunity to think about what a “photo” even is anymore — a question that’s become ever more common given things like the iPhone 14 Pro’s ramped-up sharpening and Samsung’s faked Moon photos.


You’d hope that the photo would reflect where items were when you took them, though, wouldn’t you? That seems like a basic element. Over-sharpened? Sure but it’s the same thing. Detail that isn’t there in the original? Sure, but it’s the same thing, in the same place. To me this is just a tweak too far.
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: I forgot to include a link from 2004. Thus endeth the links from the past, I guess, because otherwise we’d quickly be including links from the future, and those are much harder to find.

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.