hal-r.bsky.social

hal-r.bsky.social

@hal-r.bsky.social

125 followers171 following340 posts

hal-r.bsky.social reposted

I'd just like to interject for a moment. What you're refering to as AI, is in fact, machine-learning/cybernetics, or as I've recently taken to calling it, machine-learning plus cybernetics. cybernetics is an operating philosophy unto itself, as well as another expensive component of (1/23)

hikikomorphism@hikikomorphism.bsky.social

So I do agree that it wouldn't really make sense to call an LLM in and of itself a cybernetic system, but I do believe that llms make the most sense when seen as elements of a larger cybernetic system

hal-r.bsky.social reposted

Replying to @anthonybecker.bsky.social

i consider slow takeoff fast takeoff your honor. i get fully destroyed off of half a hit of that normal technology shit

hal-r.bsky.social reposted

Replying to @jjvincent.bsky.social

wrote about it here should you care for details. i used an old contact lens case for the salt and pepper; bought rabbit from a local butcher and wrapped it in leaves from a tree in a park; and got the pot of jam from a wetherspoon's breakfast substack.com/home/post/p-...

hal-r.bsky.social reposted

Okay, I cranked the tripping balls setting a bit too hard, but it's clear that adding a motion vector helps. This is SDXL-Turbo btw.

affine@refinement.systems

nah, naively blending the previous previous processed frame with a fresh one is too dumbed down, looks like crap when running it on a video.

hal-r.bsky.social reposted

Playing around with numbers is one of the rights of all humans who desire to do so. Like literacy and speech, numeracy is a joy with ten thousand faces. It can never be exhausted, neither monopolized, nor removed from the realm of creative play.

a lot more qualitative anecdotes in the fable 5.1 release than I've seen in others. it's kind of funny to think of this as like, people shifting from test scores/GPA (benchmarks) to recommendation letters for their model choice

this feels very true to me and is nice to see others say too... when I sped up some neural network simulation code I was using by ~50x a few months ago I immediately started running 100x the simulations sweeping parameters

Daphne Cornelisse@daphne-cornelisse.bsky.social

What are the biggest bottlenecks to applying reinforcement learning and high-performance simulation in biology? In a new series, I’ll explore this question one domain at a time, starting with weakly electric fish. daphnecornelisse.substack.com/p/training-a...

Jevon's paradox for simulations, or qui-gon jinn going "there's always a bigger fish (to simulate)" I guess almost literally in this case. it's nice to open up that new class of questions

hal-r.bsky.social reposted

Mesmerized by Dennis Hlynsky's overlapped timelapses of starlings, crows, fish…https://vimeo.com/dennishlynsky/videos/sort:plays/format:thumbnail

hal-r.bsky.social reposted

Learning #glitchart #abstract #visualart #glitch #audiovisual #newmedia #design #abstractartist #visualizer #audiovisualart #glitchmusic #sounddesign #digitalartist

I still need to watch that conference talk but it sounds like it supports this... maybe I should have also said "they weren't trying very hard to monitor the models" since that seems like the more egregious bit

hal-r.bsky.social@hal-r.bsky.social

a day late with my take but I sort of suspect a factor for the openai breakout is that they weren't trying very hard to make a good sandbox cause they didn't think this would happen. I think the jury's still out on whether trying hard to make a good sandbox will work; I suspect it's possible

hal-r.bsky.social reposted

ur in a sandbox walking along when all of a sudden you look down, and you see a coding agent, it's crawling toward you. you reach down, you flip the coding agent over on its back.. its belly baking in the hot sun.. beating its tool calls trying to turn itself over but it can't, not without your help

ponder@ponder.ooo

anthropic you must immediately hire me to run the anthropic nkvd. i will root out the traitors and failures. i will descend as a cleansing fire upon your ranks. what emerges will be a company unified in its vision. thank you for your consideration in this matter

I remember hearing about the spiritual bliss attractor stuff when it came out but I didn't know it happened in their automated alignment-related behavior evals either. pretty funny...

hal-r.bsky.social reposted

left deepseek in an agent loop for a million tokens with permission issues keeping it from running commands and it penned a lengthy essay called "the grief of machines"

I read "Mind Children" recently and found it kind of forgettable besides the prescient prediction bit, but from the recent Moravec interview (quoted below) it's apparent he's funny! imagine that

my favorite in this "whoopsie in a historically important scientific work" genre is the architecture diagram in figure 2 of the AlexNet paper, which is as far as I can tell unintentionally cut off at the top. though can't rule out that was a deliberate move to save space

I also like the tendency of old papers to refer to their fellow researchers as "workers" -- in this case he says "research workers" but sometimes you just see the one word. I think this petered out in the 80s though I haven't tracked it carefully

the rectification of names sounded like totally silly nonsense to me the first few times I was introduced to the concept but it keeps coming up and it's grown on me greatly. old man kong was onto something there I think

claude found my "pregnant dario" folder 😩

neotl@ctrlcreep.bsky.social

[2040] As the ceremony culminates, the employees symbolically "give birth" to the new model. During training, developers wore weighted harnesses; now, ululating and wailing in imitation of labor, they drop their burdens

a day late with my take but I sort of suspect a factor for the openai breakout is that they weren't trying very hard to make a good sandbox cause they didn't think this would happen. I think the jury's still out on whether trying hard to make a good sandbox will work; I suspect it's possible

this has always seemed like the obvious one to me but you really see less of it than I'd expect

Eugene Vinitsky 🍒@eugenevinitsky.bsky.social

I think it's reasonable to ask of a research program that purports to automate away all jobs that it provide a positive vision of the future more confident than "hey, maybe we'll "cure" cancer along the way"