it will definitely be rust bc the go community is pluralistic and tolerant and the rust community has an explicitly pseudo-nationalist goal of exterminating other programming languages with essentially no opposition, of course rust will do that

Watching its fellow millenials turn into unhinged boomers in real time has been eye opening

For instance, we know that strongly internalizing an essentialist identity can be harmful for people's ability to face change. Those are interesting mechanisms we can test for and potentially help reduce people's distress by changing, imho.

As always, please click through and look at the study, not just the article and certainly not just the headline, before letting this influence your worldview, but… good, fascinating study. Biggest takeaway (IMO) in the image here.

yeah that is not consent by any reasonable definition

a lot of people freaking the heck out about "someone figured out how to use AI to give them reminders to do stuff, like setting a reminder on your phone except with few-shot classifiers that can actually see what you're doing" this is called "using technology for accessibility," not "mommy issues"

It'd be nice if they'd hold *humans* to this level of standard. A one-year ban for anything even "misleading" or "biased" on arXiv? Any mistakes whatsoever? Why not apply this standard to human-created content as well?

quick reminder that copyright is morally indefensible copyright infringement is not theft it’s always ethical to redistribute copyrighted works

Big diff go brrrrr

the only meaningful point of friction i’ve found with jj is its lack of hooks. (the idea being since every file edit is immediately snapshotted into a commit there’s no way for a hook to refuse.)

if you think "female" is boring i really dont know what to tell you

llms are so much smarter than six months ago and people still refuse to recognize it

I have non-technical friends and family who've recently started using AI for their computer issues and it's really night and day for them. And I think the gap will widen over time.

Right-to-repair isn't enough, and what got lost with the RMS/free software tradition was bigger and weirder: the right to inhabit your own machine as an owner, not a tenant.

love that fast fog

the paper gestures at something like the database-of-vector-programs interpretation: “statistics extracted by the ANN much like the results of the infinite possible queries we could write in SQL” but how those extracted patterns arise and function is the thing the researchers don’t understand

doctors need to learn that respect is earned

The model uses a time-aligned context rather than one based on turns and uses alternating input and output blocks, it's still a 1D token sequence. They use two models, one for immediate interaction and one for larger tasks like web searching and tool use. They use a jointly trained, small encoder

here's a weird hypothesis: even if llms never exceed human capability directly, we'll see some increases in software quality, at least in some cases, for the following reason: excellence doesn't scale

idly considering converting my home nas box from 15 separate vms into one giant fuchsia vm running 15 components to do this i would have to invent a whole server component architecture for fuchsia since it wasnt designed for that but holy heck is its security model nice

oh hey its me

The ideal employment figure is 0%.

it's been so weird how reactionary many leftists are getting, even in anarchist camps that had been more immune than authcoms. so much antisemitism, scapegoating of intellectuals, a lot of 'retvrn' style rhetoric, and conspiracy theories all over social media, that open the door to people like her.

I’m super biased but AI tools completely changed biotech and pharma starting about 5 years ago and you’ll see the results on the market in another 5-10. It brought a whole bunch of things from impossible to possible and allowed protein engineering to become a real engineering discipline

"nonono see it's actually woke to be against vaccines and mRNA now because AI folded these proteins"

for these reasons, i believe universal income supported by taxation of economic activity is a feasible mechanism of ensuring the public welfare, that can coexist with any arbitrary arrangement of capitalist systems, and independent of what can be ad-hoc socializing particular industries or services.

something classic sci-fi shows like ST: TNG and Andor understand is that there's nothing better than scenes of meetings around a conference table

funniest part about this is the ai was 100% correct in its assessment of the situation

another problem with pressing the red button is that if the red button pressers manage to murder all the blue button pressers, you'll be stuck living in a world inhabited exclusively by red button pressers

I don't believe that ideas are permanently contaminated by their source. But since "no technical solution to social problems" is substantively a dubious claim, it seems worth remembering that its initial application was incorrect.

there's a fairly simple proof that automation will happen eventually

llms are not terribly dissimilar from human coworkers in this respect

treating it as the fallible assistant it is the best way. it still reduces work i have to do but i have to critically think about and follow up on what i get, and more so the higher the stakes of it being accurate are. also cheaper ones like in google search have lower trustworthiness but not none.

atp it is negligence to release new security-sensitive code that hasn't been audited by a frontier-class llm especially in languages that lack memory safety

historically, fact checking was part of copy editing and fact checkers did indeed contact sources as necessary to double check details — doesn’t require a full interview, just certain facts the problem is most media orgs fired all their copy editors, including fact checkers, throughout the 2010s

it’s bad on purpose to make you click

yall aren’t doing yourselves any favors reading or responding to this junk

butts

So Claude Mythos was, indeed, not marketing hype. Remember this is a general purpose model that just happens to be good at finding exploits because good models are good at lots of things. Expect similar from OpenAI & Google. And from open models in 8 months. hacks.mozilla.org/2026/05/behi...

This is the most beautiful thing I have ever seen. Looks so much prettier than endless fucking corn fields. Why aren't we in the timeline where the midwest produces more solar energy than the planet needs five times over to power billions of GPUs?

Being in favor of strong copyright protections is one of the most pro-billionaire political positions you can have

uhh,,,,,,,,,, as ai makes cognition cheaper, jevons paradox predicts there will be more demand for cognition. it does not say there will be more demand for *human* cognition! in this article screenshot, ai is fossil fuels and humans are the whalers and woodchippers www.a16z.news/p/the-ai-job...

how it feels not giving a fuck about nintendo products

though one difference from fantasy football is that if you can port Bun to Rust in 3 days for $5k worth of tokens, that upends some fundamental assumptions in my profession

realizing that chinese ai models are more likely to be running in carbon neutral datacenters

The problem is these parents just genuinely think the doctors are lying to them. It's like a mass psychosis going on. I came across a TikTok the other day of a woman who was upset that a hospital called CPS on her when she refused chemotherapy for her child with Leukemia.

another part of the problem is that doctors *do* straight up lie to you about plenty of things it’s not only a matter of misinformation about vaccines and vitamins (that is bad enough), it’s also an own goal from healthcare as a field