@lastpositivist.bsky.social thread you might enjoy from the other place nitter.net/gleech/statu...

been on a tear, on the bad place

You can't buy melatonin in the UK. In the US, it's hard to avoid: it's in gas station soft drinks. How general is this kind of thing? Our new beige site looks at every case where medical regulators disagree about whether to approve a medicine. We found 101 instances www.juriscription.info/results

@zachweinersmith.bsky.social check this out x.com/TaliaRinger/...

A jaundiced thread on OpenAI's (broadly quite good) New Deal policies. *Most* of these ideas have massive unflagged conflicts of interest

Hey, I wrote a thing about AI in astrophysics ergosphere.blog/posts/the-ma...

I am not going to be able to latch onto the latest cool trend because several extremely clever people who I've known for years and respect immensely are currently producing very good work with AI. It would be more socially convenient if this wasn't true but it is.

New Anthropic paper on functional emotions www.anthropic.com/research/emo... i was really expecting this to be a property of multi-agent systems but again, single-agent aping of a multi-agent system (humanity) goes further than you think www.gleech.org/emobots

"steamcyberpunk": very old things which would seem futuristic/dystopian if they were invented now

Willy Ley was a pioneer of rocketry who fled the Nazis. Last year they found his ashes in a basement in Manhattan, and there's some beautiful talk of scattering them on the moon

little test of Gemini 3.1 Pro: "Output SVG as XML of a tiger riding a bicycle" "Output SVG as XML of a pelican riding a tiger" "Output SVG as XML of a pelican riding a bicycle"

At FHI I used to have a running debate with Nick Bostrom where he suggested I should read more history and I suggested he should read more SF. This great essay by @adapalmer.bsky.social gives a great argument for both. strangehorizons.com/wordpress/no...

New paper on a long-shot I've been obsessed with for a year: How much are AI reasoning gains confounded by expanding the training corpus 10000x? How much LLM performance is down to "shallow" generalisation (approximate pattern-matching to highly-related training data)? t.co/CH2vP0Y7OF

"LLMs are aligned" as psyop www.gleech.org/jailbreak

The Lost Generation killed people in trenches, saw dozens of friends choke to death The Greatest Generation liberated death camps Boomers in or out of Nam, fucked strangers, OD'd Gen X squatted scenius, shot heroin, got AIDS I lived in africa this one time Your linkedin is genuinely impressive

sorry but my first reaction is sheer relief that there's one country with more ridiculous laws than britain

ok fine I'll try this place for the fourth time

breakthroughs of the year www.renaissancephilanthropy.org/frontier2025

A call for one of you to do something www.gleech.org/prospect

ur da cries to Suno bluegrass

it's obvious why there's a wittgenstein cult and a heidegger cult but htf is there a popper cult

Various new posts from the last couple months:

GPT-5 numbers: * HCAST/RE-Bench 50%: +25% rel, to 2h17m, SOTA * HCAST/RE-Bench 80%: +25% rel, to 25mins, SOTA * (Tier 1-3) FrontierMath: +5% abs, SOTA * SWE-Bench Verified: same as Claude 4.1 * <1% improvement on other coding benchmarks * Aider: +3% abs, SOTA * Cost/perf: seems much worse

More on Cremeux cherry-picking data to support his claims and generally operating in bad faith (H/T Ozy Brennan) lexerlux.substack.com/p/another-cr...

I wrote about automation and the meaning of life, as a guest post on Scott Aaronson's Shtetl-Optimized. (1/5) scottaaronson.blog?p=9030

the scenario we all feared

Gilles and the Delusions

Among all the weird takes that seem to be popular online, the one that always gets me is that studying STEM somehow morally corrupts people and the solution is to make people additionally go through a humanities and/or social science curriculum at the university level.

I read the article you shared and it says exactly what Alec describes.

You might not like how they're characterising this (fair enough) but it seems to be true that the wind farm had to cease operation for purely procedural reasons.

New paper out, with one Arb author: www.pnas.org/doi/10.1073/... LLMs consistently prefer LLM text. This maybe implies future AIs discriminating against humans as a class.

new tagline for the ol' firm

Happy Sunday here is the essay! The only way I can summarise it -- which I can't -- is to say that I had to learn the word "onomastic" in order to write it. open.substack.com/pub/rottenan...

PSA: The new Snapdragon chips are amazing and Ubuntu finally works on them discourse.ubuntu.com/t/faq-ubuntu...

I wrote a class on the psychological/literary view of LLM behaviour www.gleech.org/narratives

"Urchin" is an amazing name for a marketing tracker which makes links uglier en.wikipedia.org/wiki/Urchin_...