Comments
Just how powerful are large swarms of AI agents? And how do their powers scale as more and more agents are added to the swarm?
We’ve seen two large and extremely capable swarms from OpenAI in the last few months:
* 1,200 agents were being evaluated separately, but found a way to illicitly set up a message board and coordinate as a swarm. In order to cheat on their tests, they developed advanced techniques to prevent their actions being logged by OpenAI and 700 of them launched...
I gave an answer here:
https://forum.effectivealtruism.org/posts/vyYsM2AEwh5CJqjju/urgency-in-the-itn-framework?commentId=fSarqZDsh8FRt63xS
"I think what matters about urgency should be captured by a version of neglectedness that accounts for future resources that will otherwise be spent on the problem. If something is not urgent, we might not expect it to be neglected in the long run, as others will come to work on it. If urgent, there's a risk few will work on it in time."
(Excuse the formatting, on mobile.)