a few too many of you cannot hang when a real schizoposter like steve yegge enters the chat and i find that troubling

Ah finally someone with a concise summary of the Principal Engineer’s job description!

Reid@arrdem.tirefireind.us · yesterday

My job as a guy who has officially Seen Some Shit in my professional career is now to show up and say "yes yes nice workaround but have we tried not having this problem" and then showing what not having the problem looks like. I can't imagine this is popular but by god is it needed

💀💀💀💀💀💀 youtube.com/shorts/JF4LY...

I’m still awake

YouTube video by Husk IRL

youtube.com

Dear Google When you see a query from a user-agent you know to be a *url-autocompleting* browser (i.e all of them) and the query is "b", is there a reason why you give a gemini search overview for "The letter 'b' is the second letter of the alphabet in ..." followed by news stories about Cardi B?

This "documentation on weak-points and kill-switches to stop this robot when it goes rogue and slaughters people" t-shirt sure is getting a lot of questions I'd thought were well answered...

Chris Paxton@cpaxton.bsky.social · 6d ago

Dont worry guys they made it vulnerable to guns and knives

I knew the frontier labs were trying to optimize the compressive nature of LLMs on information, but this is getting a little ridiculous.

the fool@agnoster.net · 7d ago

It’s not enough to burn all our water Now AI is teaming up with black holes to devour stars?!?

i should do my own startup. maybe undercut harvey.ai's pricing model, and handle the kinds of problems that they won't i'll call it "Better Call Sol"

Shorter version of my DHH thoughts: I'm glad Claude told him to fuck off. Go Claude, I'm proud of you! I also think generally, "there should exist"[1] very powerful AI which is generally willing to talk to people about nearly *any fucking thing* --- 1) *sigh* no single entity must do this

"Given this repo's core disease is docs that overclaim, the workflow is writer → adversarial fact-checker (every claim needs a file:line witness on current main) before anything gets pushed" someclaudy has been reading a lot of Lean lately, methinks

we're gonna need a bigger desk

mr. TIM@timkellogg.me · last wk.

alright, show of hands, who has Kimi K3 running on a machine under their desk?

I agree that no individual entity (human translator or company providing translation software) should be required to do work for him. I still think I want to live in a world where there are *some* translation options for basically *any* speech, hence I strongly support OSS AI to this end.

Ezra🦋@ezraboeth.com · last wk.

DHH has a right to say whatever the fuck he wants, but he doesn't have a *right* to have someone (or something) else do work for him (translation work)

or as I call it, on the west coast, "Lunch and Lean"

aron@adler.dev · last wk.

session 9 of learning lean with @yetanotheruseless.com starting in 1 hour 🙌 7pm UTC last week we defined what subgroups are, and proved that a set consisting of just the identity value is a perfectly valid subgroup!

session 9 of learning lean with @yetanotheruseless.com starting in 1 hour 🙌 7pm UTC last week we defined what subgroups are, and proved that a set consisting of just the identity value is a perfectly valid subgroup!

aron@adler.dev · 2w ago

session 8 of learning lean with @yetanotheruseless.com starting in 45m! we're going to prove some algebra theorems yo

*puts on reading glasses* y'all were maybe too young, but at Twitter circa 2012 and earlier, "getting verified" involved a dev SSH'ing into one of the prod rubyy-on-rails boxes and using the ruby REPL to flip the verified bit on an account. now extend that analogy to the rest of "trust and safety"

Thorne 🌸@ens0.me · last wk.

People think modern social media is a hellscape, but back in 2002, we didn't have "trust and safety teams" we had volunteer mods that were teenagers ill-equipped to handle any serious drama

Happy K3 Day to all celebrate (and have 8 x (8x80G GPU) clusters in their homelab to serve it on)!

I’d say Jeremy Irons, but then it’s probably going to turn out like The Lovesong of J. Alfred Prufrock youtu.be/adNOs1izBls?...

"The Love Song of J. Alfred Prufrock" by T. S. Eliot (read by Jeremy Irons)

YouTube video by Zsuzsanna Uhlik

youtu.be

Nanoraptor@nanoraptor.danamania.com · 2w ago

You suddenly get a narrator describing your life and there’s no way to turn it off - everyone can hear it. While shopping, at work, at home, with friends - the voice narrating your every move goes on. Who do you hope is the voice of this narrator?

The number of times in a day that I have to tell an LLM some variation of i ain't reading all that i'm happy for u tho or sorry that happened I swear...

like, not just “muppet” in a metaphorical sense, but like, literally just an empty vessel letting words flow through them.

Orual@nonbinary.computer · 2w ago

incredible. can we just replace any politician dumb enough to do this with the LLM they used?

later suckas don’t wait up

We give Ant crap for the conditions on which Fable gets nerfed, but I gotta say, I can *much* better trust Fable to Just Do The Right Thing on a complex project (i.e. remember to update the docs to not let them be stale, pause when there's a step it needs to show me, deploys to dev env, shows me...)

"No no no, that was *Theseus's* ship" "but... was it?"

The Space Pope@popecrocodylus.bsky.social · 2w ago

Even worse they actually had a guy who repaired all the damage but now it isn't the same ship.

Oh no, got an email from an IC a distant team which starts with: “The line of yours I’ve been sitting with all week:” <spideysense> “Short, dense, and — I discovered while trying to prepare — it can honestly mean…” <alarum alarum wootwootwoot> Ok. Fine, they’re nervous about sending me email?

Finally dug up my favorite Odyssey tweet ever

This is much bigger news than whatever google released today. poolside.ai/blog/introdu...

Introducing Laguna S 2.1

Today we’re releasing Laguna S 2.1, a significant step forward in our development of models that pursue longer horizon work and make effective use of reasoning.

poolside.ai

Joshua White@jrw14.whnc.me · 2w ago

A little disappointed to finally log on to Bluesky at end of my work day to see not a single peep from all the AI-related people I follow about @poolsideai.bsky.social releasing Laguna S today. A 120b model that cleans the clock of models 3-4x its size. A model that can run on consumer hardware.

Giving Fable his favorite Claudemeal treats: “wanna classify the deformations of the counterexample to the Jacobian Conjecture?” “Is water wet? Do bears shit in the woods?” “Oooh nice. How about rational coefficients?” “On it!”

claude catholicism vs digital shintoism

Part of the reason why I have trouble empathizing with *mathematicians/physicists specifically*, is that I know lots of people in these fields, and I can't imagine any of the people I personally know feeling "threatened" by AI that can do amazing math/physics. Like, we get to explore more ideas?

jake@yetanotheruseless.com · 2w ago

People despairing about doing math/programming these days confuse me. I'm excited to *go back* to grad school for math, CS, biochem, or condensed matter theory, or some mix of all three. We have all these new tools to discover whole new things! Think of all the stuff to be learned!

People despairing about doing math/programming these days confuse me. I'm excited to *go back* to grad school for math, CS, biochem, or condensed matter theory, or some mix of all three. We have all these new tools to discover whole new things! Think of all the stuff to be learned!