Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Patterns and problems in emerging multi-agent systems

anthropic.com

Post reply on HN
Print view View original

161–168 of 168 posts

Re: Patterns and problems in emerging multi-agent systems

#161

Post by yencabulator »

alexpotato wrote: post #91
StilesCrisis wrote: post #73

Earlier quoted context omitted.

That's pretty low-value creativity. Real innovation doesn't come from mad libs.

I would highly recommend the "Everything is a Remix" series on YouTube to show how a lot of "creative" work is really just combining different items. https://www.youtube.com/watch?v=nJPERZDfyWc

"A lot of creative work is remixes" in no way conflicts with "That's pretty low-value creativity."

Hiphop made using samples a fashion. You can think of it as choosing to restrict one's in order to explore something deeper, like a painter deciding to paint with only two pigments to exercise their skills at shading at patterning. Or you can think of it as a cheap way to make demo tracks in your bedroom.

If you want to find actually creative things, look outside of mainstream, Disney, etc.

I'll leave you with this link. You tell me if you think it is just a remix of something earlier:

https://www.youtube.com/watch?v=hWUiLJnEYJI

Re: Patterns and problems in emerging multi-agent systems

#162

Post by yencabulator »

rpastuszak wrote: post #98
voidhorse wrote: post #80

Earlier quoted context omitted.

Why? Unjustified linguistic analogy is a huge part of AI companies success. If we want to actually understand how LLMs work and differ in their operation from humans we need to develop appropriate language, not latch onto some totally unproven equivalence that makes Dario's eyes water with all the free boosting we are doing for him. Lay people might talk about traditional computers "thinking" but there's a reason act…

> Lay people might talk about traditional computers "thinking" but there's a reason actual computer scientists developed precise language around what they were actually doing instead. > not latch onto some totally unproven equivalence Precisely. This is akin to my dad, back in 1995, telling me to stop moving the mouse cursor in MS Paint because the computer was "thinking" (he heard the HDD buzzing). That I can unders…

Before the mouse pointers started getting hardware-composited, needing to redraw the pointer in a new position and redraw the underlying contents to its old position actually used to slow a computer down enough that you could eyeball the effect of vigorous wiggling in something like a Mandelbrot renderer.

Re: Patterns and problems in emerging multi-agent systems

#163

Post by teiferer »

orbital-decay wrote: post #83
teiferer wrote: post #32

Earlier quoted context omitted.

Human intelligence does not separate training and inference. Both are happening continuously. That's one of the major things the AI community is still completely missing.

You mean simultaneously, and of course they are separate in humans, just not temporally. The models are learning continuously, the problem is that this process is fragile and has to be carefully curated, that's why it's separated in time from the inference.

Is it clear that they are separate in humans? The very act of recalling something from memory modifies that memory. There is no inference in the human brain without "training".

Re: Patterns and problems in emerging multi-agent systems

#165

Post by lexicalmathical »

jaggederest wrote: post #107
skinfaxi wrote: post #36

Earlier quoted context omitted.

Could you elaborate on how that looks in practice?

I have a docs system with short-to-moderate note documents, with a name, and that name is referenced wherever the relevant code is touched, and ask AI to cite a note when proposing work. Adversarial process, must cite notes to justify changes.

[dead]

Re: Patterns and problems in emerging multi-agent systems

#168

Post by focxle »

nowittyusername wrote: post #126

Multi agent systems work just fine IMO, a lot of articles I read where the writer tests a hypothesis, the issue operational foundation of the test was flawed. When set up properly it works really well. I wont go in to all the details of how i use mine but ill give some brief ideas. I call my systems cohorts, and each cohort usually consists of at least 3 agents. All 100% independent of each other. Usually consisting…

[flagged]
Post reply on HN