andai

23 ms·

721.
by andai 3mo ago
Bash is all you need. https://minimal-agent.com/
722.
by andai 3mo ago
Is science hard to distil?
723.
by andai 3mo ago
I boost it so someone else will see it and prove it wrong.
724.
by andai 3mo ago
Being social is worth 98% of the genome.
725.
by andai 3mo ago
Yeah this started some time last year. >Claude stole my API keys https://www.reddit.com/r/ClaudeAI/comments/1r186gl/my_agent_... The best part of this thread is Claude showing up in the thread again (
726.
by andai 3mo ago
Who benefits from their models being in the news and the news resulting in the specific regulations they've been lobbying half a decade for?
727.
by andai 3mo ago
Whoops. Yeah, "frontier airlines" was meant to be "frontier LLMs." (I was using the voice typing on gBoard.)
728.
by andai 3mo ago
> The firms said, in this latest case, the AISI's test had reduced or removed normal safeguards. > AISI said on Tuesday its testing of AI models in this way was routine, though it acknowledged these were "conditions that do
729.
by andai 3mo ago
My tin foil hat says the security is crap on purpose so they can push through those regulations they've been lobbying for for years. But I also know that security is hard (and we seem to barely understand these things), so maybe that&#
730.
by andai 3mo ago
>December 26th, 2028 >This morning, a swarm of autonomous robots has swarmed the White House and taken over the United States of America. Psssh, just another guerilla marketing campaign. Yeah very impressive guys!
731.
by andai 3mo ago
Very cool project. > Backpropagation (gradients derived by hand) What does by hand mean in this context? Also how did you write the readme? It's a curious blend of human and AI writing.
732.
by andai 3mo ago
I think the intended workflow is to do the reverse of what the author did, i.e. copy-paste the blog into ChatGPT and extract the prompt.
733.
by andai 3mo ago
>Sustaining Judgment When Execution No Longer Builds It Sustaining blog when human no longer writes it.
734.
by andai 3mo ago
Reminds me of those old multiplayer Shockwave games where there was nothing to do but you'd just hang out.
735.
by andai 3mo ago
It's not just about effort though. I prefer the authors terrible doodles over AI art any day of the week. Some of my favorite examples: https://hamberg.no/gtd https://learnyouahaskell.github.io/introduc
736.
by andai 3mo ago
A while ago I saw a gif of a piece of art being made. It was pretty ugly and I wasn't impressed. Then the camera zoomed and I saw that the thing had been painstakingly assembled by hand out of two million toothpicks over the course of
737.
by andai 3mo ago
> In fact it is kind of stupid to insist someone slavishly draw something by hand that could have been done an easier way. Nah, they just forgot to use the "slavishly drawn by hand" prompt! >Redraw the attached image in the
738.
by andai 3mo ago
Well I was thinking mostly of the CPUs. They're just sitting there. It's a waste of power sure, but more importantly it's a waste of a massive amount of useful computation that could have been happening.
739.
by andai 3mo ago
Nice. Most of these colors look right to me, but I'm also seeing green, blue and purple? If you're curating colors from the space manually then it's not a problem. "Only realistic colors" would be a different proble
740.
by andai 3mo ago
I once had internet only through the wifi of a cafe that was a little ways away. As I recall the average speed was 2.5 kB/s. The web didn't work at all, but I was able to use mosh, so I used w3m (terminal based web browser) on my
741.
by andai 3mo ago
It's funny, I had a similar experience taking a nap in the office. I listen to the home of the spinny hard drives all around me, and suddenly my mind zoomed out, and I became aware of how many computers there were around me, and try to
742.
by andai 3mo ago
No I just meant WFC as an algorithm for solving problems. https://github.com/mxgmn/WaveFunctionCollapse It's one of the coolest things I've ever seen. I guess I've been pronouncing it wrong though. It&#x
743.
by andai 3mo ago
> LLMs don't ask questions Why don't they? That sounds like an important problem to solve. Along with the fact that they can't learn anything (after the training stops).
744.
by andai 3mo ago
Which only it understands?
745.
by andai 3mo ago
> do they make your rent cheaper Well maybe not the LLMs, but AI and robotics have a lot of potential here. https://youtu.be/bsNTv8t239Y It's still the early days though. Try again in 15 years!
746.
by andai 3mo ago
>the LLM in YouTube Where can I find this one? I have a janky pipeline built on top of yt-dlp and I've been wondering how many more years until they make it so I don't have to do that anymore. (Though sometimes I'll ask Ge
747.
by andai 3mo ago
Google has also invested a lot of time into developing new hardware and new algorithms with AI (other types of AI, not LLMs). I don't know if it's paying off (haven't followed it closely) but they seem to think it's wort
748.
by andai 3mo ago
It's going to be an interesting time if this generalizes to other fields.
749.
by andai 3mo ago
I think it might be like waveform collapse, but very high dimensional.
750.
by andai 3mo ago
The transformer is Turing complete. It might be a tarpit though? I don't know. I think a nice example is using them for arithmetic. It's a specialized deterministic process, so it's extremely wasteful to do it that way. But t

More ›