Links (18 November 2024)
– Baldur
Bjarnason
- “Denmark: AI-powered welfare system fuels mass surveillance and risks discriminating against marginalized groups – report”. “the sweeping use of fraud detection algorithms, paired with mass surveillance practices, has led people to unwillingly –or even unknowingly– forfeit their right to privacy, and created an atmosphere of fear.” This, and not the bullshit bubble features pushed into all software, is going to be AI’s legacy. Accountability sinks that enable harms on a massive scale.
- “Google Drive Blackout in Italy After Another Major Anti-Piracy Blunder * TorrentFreak”. We’re going to see a lot more of this kind of fracturing of the web in the coming years.
- “AI is Eating Journalism, Education, and Creatives | HackerNoon”
- “Who will train tomorrow’s designers? - by Carly Ayres”. “This isn’t just about missed opportunities for individual designers. We’re witnessing the potential collapse of knowledge transfer in our industry.” Design and UX were already in bad shape, but now it’s part of a growing host of industries using “AI” as an excuse to eat its own tail.
- “A computational analysis of potential algorithmic bias on platform X during the 2024 US election | QUT ePrints”. “ The analysis reveals a structural engagement shift around mid-July 2024, suggesting platform-level changes that influenced engagement metrics for all accounts under examination. The date at which the structural break (spike) in engagement occurs coincides with Elon Musk’s formal endorsement of Donald Trump on 13th July 2024.”
- “Móðurmál receives book donation – Móðurmál”. Carrying all of those boxes out of my mother’s storage room was exhausting 😄
- “Bluesky’s success is a rejection of big tech’s operating system”. Here’s why I’m a little bit sceptical about, well, the tech industry’s sceptics. They make good points about the bubble du jour, but they also seem to have massive blind spots. I’m on Bluesky (and Mastodon) but anybody choosing Bluesky isn’t rejecting anything. They’re just replacing a broken mousetrap with a functioning one.
- “GitHub - 11ty/eleventy-import: Utility to import content from multiple services (and a CLI, too)”. This looks useful.
- “Human rights in Dubai - Wikipedia”. A writer who occasionally writes interesting things about tech and business wrote a lengthy adulation of Dubai over the weekend so instead of linking to that, I’m linking to the page on human rights in Dubai instead.
- “Biden Asked Microsoft to “Raise the Bar on Cybersecurity.” He May Have Helped Create an Illegal Monopoly. — ProPublica”. “When the White House welcomed Microsoft’s offer of $150 million in tech services, it helped the world’s largest software provider tighten its grip on federal business and freeze out competitors.”
- “Escape from Twitter: The Future of Social Media Is Decentralized”
- “AI is the enshittification of recruitment | Tales about Software Engineering”. “ In my opinion, if you are going to use LLMs to write your CVs, you are sailing pretty close to the wind. I wouldn’t exactly call it dishonest or cheating, but a CV is supposed to put your best foot forward and sell yourself. Are you really selling yourself if you can’t be bothered to write a few lines about yourself?”
- “Apple study exposes deep cracks in LLMs’ “reasoning” capabilities - Ars Technica”
- “Unmanaged Climate Risks Undercut AI’s Investment Thesis”
- “Muhammad Ali’s Photographer Wins $1.65 Million in Copyright Lawsuit | PetaPixel”
- “Researchers sound alarm on dual-use AI for defense - Defense One”. “ But a study out today from a group of prominent AI scholars argues that dual-use AI tools would increase the odds of innocent civilians becoming targets due to bad data—and that the tools could easily be gamed by adversaries.”
- “AI overwhelmingly prefers white and male job candidates in new test of resume-screening bias – GeekWire”
- “Why Hollywood Workers Feel Stuck: Battle for Opportunity”
- ““It’s Not About Making the Word Count, But Making the Work Count” • Buttondown”
- “2024 is set to be the first year that breaches the 1.5°C warming limit | New Scientist”
- “Trump Threatens New York Times, Penguin Random House over Critical Coverage - Columbia Journalism Review”. “Legal letter follows complaints aimed at CBS News, the Washington Post, and the Daily Beast.”
- “Denzel Washington Says He Had a Gay Kiss In ‘Gladiator 2’ But It Was Cut - Gayety”
- “‘No sign’ of promised fossil fuel transition as emissions hit new high | Cop29 | The Guardian”. “ Despite nations’ pledges at Cop28 a year ago, the burning of coal, oil and gas continued to rise in 2024”
- “When to use CSS text-wrap: balance; vs text-wrap: pretty; - Stephanie Stimac’s Blog”
- “Digital Infrastructure and Why Open Source Is Not Enough Anymore”. “ There’s a company that does weird shit, spends money in weird ways that gets a lot of outrage in the community, like what the fuck are you even doing? But do we really have any infrastructure that could take over?”
- “Bret Victor - De Programmatica Ipsum”
- “What if AI doesn’t just keep getting better forever? - Ars Technica”. “ But if current LLM training methods are starting to plateau, the next big breakthrough might come via specialization. Microsoft, for one, has already shown some success with so-called small language models that focus on specific types of tasks and problems.” I find it incredibly fucking annoying that this was exactly my core recommendation in my book 18 months ago but nobody fucking listened.
- “Not even Spotify is safe from AI slop - The Verge”. “ But when a fake album was posted on September 26th, it didn’t budge. Mena emailed Spotify to tell them there’d been a mistake. The streamer responded two weeks later, on October 8th: “It looks like the content is mapped correctly to the artist’s page. If you require further assistance, please contact your music provider. Please do not reply to this message.” As of November 8th, the fake Standards album was still right there under the band’s verified, blue-checked name. It was finally removed by November 11th.”
- “Reading and Writing as Human Expression & Connection - Jim Nielsen’s Blog”
- “TBM 321: “Reducing Complexity” - by John Cutler”
- “Two books, no longer apart. — Ethan Marcotte”
- “You have agency. Use it. - by Suw Charman-Anderson”. “ One of the reasons that grift is so effective is that writers have so few legitimate options to progress their career. There is no career ladder to climb. You can’t get a job as an entry level writer and then get a promotion at your annual review. Success doesn’t lead to more success.” The podcast she links to also makes for a good listen
- “‘No sign’ of promised fossil fuel transition as emissions hit new high | Cop29 | The Guardian”. “ Despite nations’ pledges at Cop28 a year ago, the burning of coal, oil and gas continued to rise in 2024”. But suuuure your use of LLMs and diffusion models magically don’t cause any emissions and in no way mean that you’re a complete asshole
- “I’m a neurology ICU nurse. The creep of AI in our hospitals terrifies me - Coda Story”. “ We felt the system was designed to take decision-making power away from nurses at the bedside.” This is a classic case of using AI as an accountability sink and I think it’s likely that this, not AI for office work, is the future of this tech. Accountability sinks designed to enable human misery on a national or even global scale.