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...
What is EA's stance on Gain-of-Function Research? (OC)
I recently interviewed a Columbia University parasitologist on GOF research. I picked him, in part, because he was strongly in favor of conducting GOF. According to him, the risks are minimal, and the benefits far outweigh them.
I'm not a scientist (obviously), but I can't say I agree with the entirety of his argument on the risks. Smallpox has been released from the lab 3 times, so we know that human error has allowed a deadly virus to escape from a lab and kill people.
It's pretty obvious that GOF has a lot of upsides though. Many scientists studying it are conducting this research explicitly to prevent pandemics in the future.
It doesn't seem obvious to me which route will produce the least risk.
Maybe we should send all scientists doing this kind of research into a quarantine bubble, and have them do 6 weeks on/6 weeks off shifts. Other than that, I'm out of ideas.
Clip of the interview if you're interested: