A new smartphone-like device will analyse skin lesions and refer them to a dermatologist to see if they’re cancerous. CC-licensed photo by Sue Thompson 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 9 links for you. More suntan lotion! I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.
Hair sample that put a man in prison turned out to be dog hair • Reason
Lenore Skenazy:
»
In 2009, the National Research Council published a report about forensic science “and that really marked a turning point in the birth of the forensic reform movement,” says [executive director of the Wisconsin-based Center for Integrity in Forensic Sciences (CIFS), Kate] Judson. “It was the first time a big government agency brought together scientific experts and legal experts and said, ‘Let’s talk about what they’re saying in court and whether it passes scientific muster.'”
One big topic was hair samples. A hair with its root can provide actual DNA evidence, says Judson. But other comparisons of color and texture, even examined under the microscope, have led to many wrongful convictions. She points to the case of Santae Tribble, convicted of murder at 17, despite evidence that he had been elsewhere when the crime occurred.
An FBI analyst at his trial testified that there was just a one in 10 million chance that the hair found on a stocking mask at the crime scene belonged to someone other than Tribble.
But after spending over 20 years in prison, Tribble was cleared when the hairs were retested and none of them matched. (At least one was dog hair.)
«
Plenty more examples of miscarriages of justice which leave you wondering quite how incompetent or malicious the people involved in the prosecution must have been.
unique link to this extract
The first AI medical device that can detect all major skin cancers just received FDA approval • DigiAlps LTD
»
On January 17, 2024, the US Food and Drug Administration (FDA) announced it had cleared DermaSensor as the first AI-powered medical device able to detect the three most common types of skin cancer: melanoma, basal cell carcinoma, and squamous cell carcinoma. Developed by medical technology company DermaSensor, Inc., the device uses a non-invasive light-based technology called elastic scattering spectroscopy (ESS) to analyze suspicious lesions on a cellular level and provide real-time cancer risk assessments to guide physicians. According to DermaSensor’s website, the device will be priced through a subscription model at $199 monthly for five patients or $399 monthly for unlimited use.
DermaSensor looks similar to a smartphone with a pointed tip on the bottom that is used to scan skin lesions. When the tip touches the skin, it projects different wavelengths of light that penetrate the skin and interact with cells. Healthy cells absorb and reflect light differently than cancerous cells based on changes at the subcellular level. An integrated AI model can analyze these light interaction patterns and identify characteristics that can indicate the presence of melanoma, basal cell carcinoma, or squamous cell carcinoma.
«
Ah, but to clarify: “The device does not make a formal cancer diagnosis but flags suspicious lesions for closer evaluation by a dermatologist.” In other words, we leave the really tricky stuff to humans, as previously.
unique link to this extract
FTC orders Intuit to cease “deceptive” turbotax ads • ProPublica
Justin Elliott and Paul Kiel:
»
The Federal Trade Commission has ordered the maker of TurboTax to stop what it called years of widespread deceptive advertising for “free” tax-filing software.
The order, released Monday, was accompanied by a 93-page opinion that harshly criticized Intuit, the Silicon Valley company behind TurboTax. Intuit’s “deceptive ad campaign has been sufficiently broad, enduring, and willful to support the need for a cease-and-desist order,” the commission’s opinion stated.
The order caps off a process that started four years ago when the FTC launched an investigation in response to a series of ProPublica stories documenting Intuit’s ad tactics. ProPublica revealed how millions of Americans were lured into paid tax preparation products even though they were eligible to file for free through a government-sponsored program. Huge sums of money are at stake: In a single year, tax prep companies led by Intuit generated $1bn in revenue from customers who should have been able to file for free, according to one analysis.
In a statement, Intuit said it planned to appeal the order in federal court. “There is no monetary penalty in the FTC’s order, and Intuit expects no significant impact to its business,” the statement said, adding that the company “has always been clear, fair, and transparent with its customers.”
Sam Levine, the director of the FTC’s Bureau of Consumer Protection, said in a statement that the order was intended to send a message to all companies: “‘free’ means free — not ‘free for a few’ or ‘free for some.’ Businesses can expect an FTC enforcement action if they harness the power of ‘free’ in the dishonest way Intuit did.”
«
Intuit’s misleading advertising around tax filings in the US have been legendary. Great to see that it’s finally being stopped.
unique link to this extract
Raworth on the Run: ‘My joints are still capable of carrying me along’ • Runners World
Sophie Raworth:
»
I am lying in an MRI scanner in London. For an hour and a half, I’m not allowed to move. The machine whirs as images are taken of every part of my knees, hips and core muscles around my pelvis. ‘We are slicing you up like salami with our images,’ I am told. I’ve chosen to come here in the name of research. But ever since I said yes six weeks ago, I’ve been stricken by fear, convinced that I’m going to be told that running 1,500 miles a year in my fifties is damaging my ageing joints. ‘What about your knees?’ is the question runners get asked all the time. Since I took up running more than a decade ago, my knees have always felt fine. I’ve had my fair share of injuries. But my bones and joints have always felt like they’re coping with the load. Recently, though, a strange crackling noise has begun emanating from my knees when I go upstairs. I’ve convinced myself it’s the running. Lying inside this machine, I’m bracing myself for bad news.
The man who is about to determine my fate with his scans is Alister Hart, an orthopaedic surgeon who, for the past six years, has been carrying out pioneering research into what running actually does to your hips and knees. His interest was sparked after he ran his first marathon a decade ago at the age of 42. For days afterwards, he limped around the hospital where he works. ‘It really got me thinking about what it was doing to my body,’ he says.
And so he began the largest and most detailed study of the knees of middle-aged marathon runners to date. He did MRI scans of 164 knees belonging to 82 runners, all in their forties, all first-time marathon runners with no known injuries. He wasn’t sure what he would find. ‘There was a risk we could have killed off running,’ he says, smiling. The results surprised him. The runners were scanned before and after a four-month training plan that culminated in the London Marathon. ‘It looks like running, even up to marathon level, is actually good for the bones. It strengthens them. And this, in turn, is good for the cartilage; we really didn’t expect that at all.’
«
Yes, it’s Raworth, the BBC newsreader and interviewer. She’s remarkably quick. Has done the Marathon des sables. And her knees make that noise too.
unique link to this extract
Soon will come a day that none of this exists • The Discourse Blog
Jack Crosbie:
»
It certainly seems like publications as we know them are, unless something drastically changes to the US tax code, going kaput. They dead. Some of them will cling on for a while and the brand names may persist, but in general, the idea of an organized institution where a group of journalists all collaboratively publish under one masthead in service of, generally, a shared ideological goal—that’s going tits up. There are only so many blogs a really good gochujang cookie recipe can buy, and sooner or later even the Grayest of Ladies are going to hit that limit.
This presents a problem, in my opinion. The good thing about publications is that with enough people working at them and enough people reading them you get something that approaches popular accountability. These publications are widely read and powerful enough that the people who run them were generally incentivized not to fuck up in public and embarrassing ways; couple that with a tiny shred of public altruism and belief in “the truth,” or at least in journalism as a foil to institutional power in other governments and industries, and you get for better or worse a mostly free press that mostly does good work that makes society in general a bit better off. That’s the best we can really hope for, I think.
But right now you basically can’t make money doing that, and nobody who has money is willing to support policy that would shuffle around the money to let that kind of industry exist (at this point, it would take a strong federal government hacking that money out of the tech monopolies and big financial institutions’ bottom lines, which does not seem likely).
What we’re going to get, then, is a whole lot more blogs like this. The one you’re reading. Discourse Blog. “Great!” you may think. “Independent, compassionate, reader-supported journalism, that also publishes funny stuff about birds!” First off, thank you, we know. We’re great. But the problem is there are many many people under this same model who are not great
«
And since we’re talking about the death of formal media..
unique link to this extract
L.A. Times to lay off at least 115 people in the newsroom • Los Angeles Times
Meg James, “senior entertainment writer”:
»
The cuts were necessary because the paper could no longer lose $30m to $40m a year without making progress toward building higher readership that would bring in advertising and subscriptions to sustain the organization, the paper’s owner, Dr. Patrick Soon-Shiong, said Tuesday.
Drastic changes were needed, he said, including installing new leaders who would focus on strengthening the outlet’s journalism to become indispensable to more readers.
“Today’s decision is painful for all, but it is imperative that we act urgently and take steps to build a sustainable and thriving paper for the next generation. We are committed to doing so,” Soon-Shiong said.
Senior editors were part of the purge, including Washington bureau chief Kimbriell Kelly, deputy Washington bureau chief Nick Baumann, business editor Jeff Bercovici, books editor Boris Kachka, and music editor Craig Marks. The Washington bureau, photography and sports departments saw dramatic cuts, including several award-winning photographers. The video unit was hollowed out.
«
The 115 layoffs are about 20% of the newsroom. This is the entrance hall to the death spiral: there simply isn’t a place where a newspaper focused on a city will break even, especially after getting rid of so many senior staff. I have the feeling that Meg James was the only person left to write the story.
unique link to this extract
Google introduces new Circle to Search feature on Android • Google Blog
Cathy Edwards is VP and GM of Seach:
»
Our phones are a window to the world’s information — whether it’s to explore a passion, solve a problem, buy the perfect gift, learn a new skill or simply find a reason to smile. And when you’re truly immersed in a moment of discovery or exploration, it can feel disruptive to stop what you’re doing and switch to another app to learn more.
That’s why we’re introducing Circle to Search, a new way to search anything on your Android phone without switching apps. Now, with a simple gesture, you can select what you’re curious about in whatever way comes naturally to you — like circling, highlighting, scribbling or tapping — and get more information right where you are.
For example, maybe you need help identifying a few items a creator wore in their “Outfit of the Day” video, but they didn’t tag the brands. Just long press the home button or navigation bar on your Android phone to activate Circle to Search. From there, you can select any item that you see with your preferred gesture — like circling their sunglasses — to quickly find similar, shoppable options from retailers across the web. You could scribble the bag and tap on the boots to look those up, too — all without leaving where you are. When you’re done, simply swipe away and you’re right back where you started.
Other times, you might want to ask more complex questions about what you see. Let’s say you’re browsing social media and come across an image of a delicious corn dog with some unique toppings. You’ve been seeing a lot of these online and are curious to learn more.
With multisearch — the ability to search with text and images at the same time — and our latest AI-powered upgrades, you can more easily understand concepts, ideas or topics from helpful information that’s pulled together from across the web. Simply circle the corn dog and ask a question, like “Why are these so popular?”
«
This is rather neat – though one does have to wonder how accurate it will be. What’s an acceptable error rate? 5%? 10%?
unique link to this extract
Google CEO tells employees to expect more job cuts in 2024 • The Verge
Alex Heath:
»
Google has laid off over a thousand employees across various departments since January 10th. CEO Sundar Pichai’s message is to brace for more cuts.
“We have ambitious goals and will be investing in our big priorities this year,” Pichai told all Google employees on Wednesday in an internal memo that was shared with me. “The reality is that to create the capacity for this investment, we have to make tough choices.”
So far, those “tough choices” have included layoffs and reorganizations in Google’s hardware, ad sales, search, shopping, maps, policy, core engineering, and YouTube teams.
“These role eliminations are not at the scale of last year’s reductions, and will not touch every team,” Pichai wrote in his memo — a reference to when Google cut 12,000 jobs this time last year. “But I know it’s very difficult to see colleagues and teams impacted.”
«
Why the job growth and then the job shrinkage? The implication of the second paragraph is that it’s to fund the AI work, but does that really require all those layoffs? Google begins to look like Just Another Company.
unique link to this extract
A ‘shocking’ amount of the web is already AI-translated trash, scientists say • Vice
Jules Roscoe:
»
“We actually got interested in this topic because several colleagues who work in MT [machine translation] and are native speakers of low resource languages noted that much of the internet in their native language appeared to be MT generated,” Mehak Dhaliwal, a former applied science intern at AWS and current PhD student at the University of California, Santa Barbara, told Motherboard. “So the insight really came from the low-resource language speakers, and we did the study to understand the issue better and see how widespread it was.”
“With that said, everyone should be cognizant that content they view on the web may have been generated by a machine,” Dhaliwal added.
The study, which was submitted to the pre-print server arXiv last Thursday, generated a corpus of 6.38 billion sentences scraped from the web. It looked at patterns of multi-way parallelism, which describes sets of sentences that are direct translations of one another in three or more languages. It found that most of the internet is translated, as 57.1% of the sentences in the corpus were multi-way parallel in at least three languages.
Like all machine learning efforts, machine translation is impacted by human bias, and skews toward languages spoken in the Western world and the Global North. Because of this, the quality of the translations varies wildly, with “low-resource” languages from places like Africa having insufficient training data to produce accurate text.
“In general, we observed that most languages tend to have parallel data in the highest-resource languages,” Dhaliwal told Motherboard in an email. “Sentences are more likely to have translations in French than a low resource language, simply by virtue of there being much more data in French than a low resource language.”
«
Two things: the paper isn’t peer-reviewed; this is machine translation, not machine generation.
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 