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 downvoted this post. I watched the first hour of the video and was very unimpressed by the "argument" in it. It seems to be a mix of implicit conspiracism, irrelevant tangents, and intro philosophy of science.
It does (correctly) point out that the replication crisis revealed many weaknesses in the way science has been conducted, but the discussion is superficial. And whereas most scientists who learn about the replication crisis advocate for greater rigor (e.g. larger sample sizes, more diverse samples, preregistration), the video implies that the real problem is that scientists have been making some unwarranted metaphysical/ontological assumptions. For example, scientists should be more open to the idea that extra sensory perception is real??
I think a better use of time would be reading Stuart Ritchie's book Science Fictions, which more clearly and cogently discusses the replication crisis and problems in science more generally.
I understand that you didn't like the way he speaks, but giving down vote without watching the whole thing seems to be bit unfair. In second part he departs from talking about replication crisis and dives deeper into the topic. I'd recommend watching it full before giving it a review.