Even job applications are being infested by AI-generated applications – which is leading to AI-based rejections. Will it mean AI-based jobs? CC-licensed photo by Loco Steve 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 9 links for you. Applicable. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.
All the little data • New Cartographies
Nicholas Carr:
»
Even my consumption of cultural goods—an ugly phrase, yes, but it seems apt—is shadowed by metadata. When the graphical user interface was introduced to personal computers in the early 1980s, the scroll bar habituated us to a visual indicator of our progress through a document. Now, pretty much all viewing, listening, and reading is tracked, visually or numerically, in real time. When I’m listening to a song, a glance at the progress bar tells me, to the second, how much time has elapsed since the tune began and how much remains before it ends. The same goes for TV shows and movies and videos.
When I’m reading an ebook, I’m kept apprised of the percentage of the text I’ve made it through. When I’m looking over the homepage of a newspaper or magazine site, I’m told how long it will take to read each article. Here’s a “3 min read.” There’s a “7 min read.” (This essay, for the record, is a thirteen-minute read, and you have nine minutes to go.) Every photo on my phone offers its own little data dump: where and when it was taken, the aperture and ISO settings, the exposure time, the image’s size in pixels and bits. My pictures tend to be amateurish, but the data always looks professional.
«
A lovely little soliloquy about how much data we have raining down on us, and yet how little it can truly tell us. The computer records all the photo’s metadata, but only we know if it’s a good photo, or one with meaning for us. The data shows what time we had a phone call and how long it lasted; only we know if it was important or trivial (and even the duration might not tell you that).
How can you capture that? You can’t.
unique link to this extract
The companies that help people vanish • BBC Worklife
Bryan Lufkin:
»
In Japan, these people are sometimes referred to as “jouhatsu”. That’s the Japanese word for “evaporation”, but it also refers to people who vanish on purpose into thin air, and continue to conceal their whereabouts – potentially for years, even decades.
“I got fed up with human relationships. I took a small suitcase and disappeared,” says 42-year-old Sugimoto, who’s just going by his family name for this story. “I just kind of escaped.” He says that back in his small hometown, everybody knew him because of his family and their prominent local business, which Sugimoto was expected to carry on. But having that role foisted upon him caused him such distress that he abruptly left town forever and told no one where he was going.
From inescapable debt to loveless marriages, the motivations that push jouhatsu to “evaporate” can vary. Regardless of their reasons, they turn to companies that help them through the process. These operations are called “night moving” services, a nod to the secretive nature of becoming a jouhatsu. They help people who want to disappear discreetly remove themselves from their lives, and can provide lodging for them in secret whereabouts.
“Normally, the reason for moving is something positive, like entering university, getting a new job or a marriage. But there’s also sad moving – for example, like dropping out of university, losing a job or escaping from a stalker,” says Sho Hatori, who founded a night-moving company in the 90s when Japan’s economic bubble burst. At first, he thought financial ruin would be the only thing driving people to flee their troubled lives, but he soon found there were “social reasons”, too. “What we did was support people to start a second life,” he says.
Sociologist Hiroki Nakamori has been researching jouhatsu for more than a decade. He says the term ‘jouhatsu’ first started being used to describe people who decided to go missing back in the 60s. Divorce rates were (and still are) very low in Japan, so some people decided it was easier to just up and leave their spouses instead of going through elaborate, formal divorce proceedings.
“In Japan, it’s just easier to evaporate,” says Nakamori. Privacy is fiercely protected: missing people can freely withdraw money from ATMs without being flagged, and their family members can’t access security videos that might have captured their loved one on the run.
«
This piece also available as a BBC video and as an audio episode.
unique link to this extract
AI sludge has entered the job search • The New York Times
Sarah Kessler:
»
Katie Tanner, a human resource consultant in Utah, knew the job would be popular: It was fully remote, was at a tech company and required only three years of experience.
But she was still shocked by the response on LinkedIn. After 12 hours, 400 applications had been submitted. By 24, there were 600. A few days later, there were more than 1,200, at which point she removed the post. Three months later, she’s still whittling down candidates.
“It’s crazy,” she said. “You just get inundated.”
The number of applications submitted on LinkedIn has surged more than 45% in the past year. The platform is clocking an average of 11,000 applications per minute, and generative artificial intelligence tools are contributing to the deluge.
With a simple prompt, ChatGPT, the chatbot developed by OpenAI, will insert every keyword from a job description into a résumé. Some candidates are going a step further, paying for A.I. agents that can autonomously find jobs and apply on their behalf. Recruiters say it’s getting harder to tell who is genuinely qualified or interested, and many of the résumés look suspiciously similar.
“It’s an ‘applicant tsunami’ that’s just going to get bigger,” said Hung Lee, a former recruiter who writes a widely read newsletter about the industry.
One popular method for navigating the surge? Automatic chat or video interviews, sometimes conducted by A.I. Chipotle’s chief executive, Scott Boatwright, said at a conference this month that its A.I. chatbot screening and scheduling tool (named Ava Cado) had reduced hiring time by 75 percent.
HireVue, a popular A.I. video interview platform, offers recruiters an option to have A.I. assess responses and rank candidates.
But candidates can also use A.I. to cheat in these interviews, and some companies have added more automated skill assessments early in the hiring process.
«
But it’s fine because an AI will take the job anyway.
unique link to this extract
Why I am no longer an AI Doomer • The Deep Dish
Richard Meadows:
»
Intelligence is the ability to solve a broad range of problems. ChatGPT is very intelligent: it can solve a broad range of problems across many domains.
Your old Casio pocket calculator is also good at solving problems— albeit within a narrower domain— but it feels weird to call it ‘intelligent’. What’s missing?
To help get at the thing we’re actually interested in, the usual move is to add the concept of agency. We don’t just solve problems; we solve them in the pursuit of our goals (unlike a chatbot or calculator, which just sits around idly until a human gives it something to do). An agent doesn’t just think; it thinks for itself. As Sarah Constantin puts it in her very-good essay of a similar name, agency is what makes humans so powerful—and so dangerous.
But we actually get a lot more clarity when we keep these two concepts separate. Intelligence does not require agency: your pocket calculator is capable of instantly solving problems that very few humans could tackle in their heads (or at all). And agency does not require much in the way of intelligence: an amoeba is an agent in the world, but it is only a tiny bit smarter than a rock.
The third element that needs to be teased apart is creativity, defined by David Deutsch as the ability to come up with new explanatory knowledge; this being a contender for the secret sauce that separates humans from other very smart animals (and from current-level AIs). Crucially, this is a step-change rather than a spectrum of ability: if you can explain something, you can in principle explain anything that is explicable.
…A true AGI will necessarily be an agent, with its own desires, whims, and goals. And a true AGI will necessarily be creative, in the Deutschian sense: it will be able to create new explanatory knowledge.
Current-level AI has neither of these properties, and has no prospect of attaining them via current approaches. It’s incredibly smart, but it’s still much more like a pocket calculator than it is like a person.
«
A good explanation; there’s more about the gap between this and “life”.
unique link to this extract
Sam Altman says GPT-5 coming this summer, open to ads on ChatGPT—with a catch • AdWeek
Trishla Ostwal:
»
OpenAI CEO Sam Altman announced on a new company podcast today that GPT-5 is expected to launch this summer, marking the next major leap in the company’s generative AI capabilities. However, he did not disclose a specific date.
The announcement comes amid rising competition in the AI arena and growing scrutiny over how these tools are developed and deployed. Business Insider reported that GPT-5 is shaping up to be a significant upgrade over GPT-4, with early testers calling it “materially better.”
Meanwhile, OpenAI’s primary revenue comes from enterprise customers buying beefed-up versions of ChatGPT, and GPT-5 is poised to be the next big play to sustain that momentum.
Altman also weighed in on the possibility of ads on ChatGPT, saying he’s “not totally against” the idea—a shift that could reshape how the chatbot is monetized.
However, he warned that it would take “a lot of care” to get the experience right. Unlike social media or web search, where users expect some level of monetization, Altman emphasized that modifying the model’s output based on who pays for the ad would be “a trust-destroying moment” for its users.
Instead, he floated the idea of showing ads outside the large language model’s output stream. Altman didn’t specify what form those ads might take or where they might appear, such as a sidebar or footer.
“But the burden of proof there, I think, would have to be very high,” he said. “And it would have to feel really useful to users and really clear that it was not messing with the LLM’s output.”
«
I’m old enough to remember when Google was absolutely insistent that it would never put ads on its homepage (the one before you’ve carried out a search). It did indeed turn down a very large sum from a credit card company wanting just that in its early years. But in its later years it wasn’t above advertising its own products, particularly Chrome. After all, every user who switched from Safari or Firefox to Chrome both saved Google money on search clicks, and increased the amount of user data it could collect.
All of which is to say that people will look at the ads being pulled into OpenAI as a goldmine about what people want.
unique link to this extract
Websites are tracking you via browser fingerprinting • Texas A&M University Engineering
»
New research led by Texas A&M University found that websites are covertly using browser fingerprinting — a method to uniquely identify a web browser — to track people across browser sessions and sites.
“Fingerprinting has always been a concern in the privacy community, but until now, we had no hard proof that it was actually being used to track users,” said Dr. Nitesh Saxena, cybersecurity researcher, professor of computer science and engineering and associate director of the Global Cyber Research Institute at Texas A&M. “Our work helps close that gap.”
When you visit a website, your browser shares a surprising amount of information, like your screen resolution, time zone, device model and more. When combined, these details create a “fingerprint” that’s often unique to your browser. Unlike cookies — which users can delete or block — fingerprinting is much harder to detect or prevent. Most users have no idea it’s happening, and even privacy-focused browsers struggle to fully block it.
“Think of it as a digital signature you didn’t know you were leaving behind,” explained co-author Zengrui Liu, a former doctoral student in Saxena’s lab. “You may look anonymous, but your device or browser gives you away.”
This research marks a turning point in how computer scientists understand the real-world use of browser fingerprinting by connecting it with the use of ads.
“While prior works have studied browser fingerprinting and its usage on different websites, ours is the first to correlate browser fingerprints and ad behaviors, essentially establishing the relationship between web tracking and fingerprinting,” said co-author Dr. Yinzhi Cao, associate professor of computer science and technical director of the Information Security Institute at Johns Hopkins University.
«
The irresistible force online: advertising.
unique link to this extract
How OpenElections uses LLMs • The Scoop
Derek Willis:
»
In the 12-plus years that we’ve been turning official precinct election results into data at OpenElections, the single biggest problem has been converting pictures of results into CSV files. Many of the precinct results files we get are image PDFs, and for those there are essentially two options: data entry or Optical Character Recognition. The former has some advantages, but not many.
While most people are not great at manual repetitive tasks, you can improve with lots of practice, to the point where the results are very accurate. In the past we did pay for data entry services, and while we developed working relationships with two individuals in particular, the results almost always contained some mistakes and the cost could run into the hundreds of dollars pretty quickly. For a volunteer project, it just didn’t make sense.
We also used commercial OCR software, most often Able2Extract, which did pretty well, but had a harder time with PDFs that had markings or were otherwise difficult to parse. Thankfully, most election results PDFs are in one of a small handful of formats, which makes things a bit less complicated, but commercial OCR has too many restrictions.
For parsing image PDFs into CSV files, Google’s Gemini is my model of choice, for two main reasons. First, the results are usually very, very accurate (with a few caveats I’ll detail below), and second, Gemini’s large context window means it’s possible to work with PDF files that can be multiple MBs in size.
…Speed isn’t the most important factor here, though: accuracy is, and using LLMs still means a system of checks to ensure that the results are what the originals say they are. One step in that is taken care of by a suite of tests that run every time a new or changed CSV gets pushed to one of our data repositories. Those tests look for some formatting issues, duplicate records and basic math inconsistencies. A second step – for now manual – is verifying that multiple totals derived from the precinct CSV match the numbers in an official cumulative report.
«
I haven’t tried formatting a PDF into a CSV (the ones I deal with tend to be small, and not for stuffing into systems which are retaining their logs for court hearings), but I’d like an explanation of why LLMs should be good at reading PDFs.
unique link to this extract
Ear wax as a possible screening medium for Parkinson’s disease • American Chemical Society
»
Most treatments for Parkinson’s disease (PD) only slow disease progression. Early intervention for the neurological disease that worsens over time is therefore critical to optimize care, but that requires early diagnosis. Current tests, like clinical rating scales and neural imaging, can be subjective and costly. Now, researchers in ACS’ Analytical Chemistry report the initial development of a system that inexpensively screens for PD from the odours in a person’s ear wax.
Previous research has shown that changes in sebum, an oily substance secreted by the skin, could help identify people with PD. Specifically, sebum from people with PD may have a characteristic smell because volatile organic compounds (VOCs) released by sebum are altered by disease progression — including neurodegeneration, systemic inflammation and oxidative stress. However, when sebum on the skin is exposed to environmental factors like air pollution and humidity, its composition can be altered, making it an unreliable testing medium. But the skin inside the ear canal is kept away from the elements. So, Hao Dong, Danhua Zhu and colleagues wanted to focus their PD screening efforts on ear wax, which mostly consists of sebum and is easily sampled.
To identify potential VOCs related to PD in ear wax, the researchers swabbed the ear canals of 209 human subjects (108 of whom were diagnosed with PD). They analyzed the collected secretions using gas chromatography and mass spectrometry techniques. Four of the VOCs the researchers found in ear wax from people with PD were significantly different than the ear wax from people without the disease.
«
There: you were not expecting to discover this when you woke up today. Of note: dogs sniff each others’ ears too when they meet – so perhaps they know some version of this too.
Global sea levels rising twice as fast as they did last century, according to major scientific report • Sky News
Victoria Seabrook:
»
Global sea levels are now rising twice as fast as they did last century, according to a major new scientific report.
The study – which takes a laser focus on climate change in the 2020s, a critical decade to stop the worst damage – finds all 10 measures are going in the wrong direction.
And most of them are doing so at a faster rate.
The findings are “unprecedented” but “unsurprising”, given the world continues to pump record levels of planet-warming gases into the atmosphere.
“We see a clear and consistent picture that things are getting worse,” said lead author Professor Piers Forster.
However, the rate by which emissions are increasing has slowed down, offering a ray of hope they will soon reach their peak.
The new study found sea levels are now rising on average twice as fast, at 4.3mm a year on average since 2019, up from 1.8mm a year at the turn of the 20th century.
The acceleration is stark, but within the realms of what scientists expected. That’s because the warming atmosphere has sent more melting ice flowing into the sea, and the ocean water expands as it warms.
For the island nation of the UK, which risks coastal flooding, cliff falls and damage to homes and buildings, with 100,000 properties expected to be threatened with coastal erosion in England within 50 years.
«
| • 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 