I’m not worried about CHAI as a whole getting ‘distracted’, at least in the next couple of years. There are several people still at CHAI who are very definitively focused on existential risk—Andrew Critch, Adam Gleave, and Daniel Filan, for example. It’s also led by Stuart Russell, who is probably the figure most associated with AI x-risk research amongst the AI research community, with perhaps the exception of Nick Bostrom.
If you think that I’m better at reducing x-risk than they are (e.g. because my beliefs are better), then that seems like a plausible reason to worry. I do expect that the research that CHAI does will be somewhat different now that I am not there; I have many disagreements with many people at CHAI. But I wouldn’t worry about them being distracted.
I personally donated to CHAI this year; part of my reasoning was that CHAI could use unrestricted funding for research engineering roles. (Note that I am now at DeepMind, so I was less worried about conflicts of interest, though obviously I still have many friends at CHAI.)
Poor Quality Research
I wanted to note I agree with most of this section, and have adopted a similar policy. I do think there are more downsides to calling out poor research—the main one I’d note is that in general because communication is hard people (or at least I) tend to be systematically too negative about other people’s research. (See previous link for more details.)
Re: rot13 and donation decisions:
I’m not worried about CHAI as a whole getting ‘distracted’, at least in the next couple of years. There are several people still at CHAI who are very definitively focused on existential risk—Andrew Critch, Adam Gleave, and Daniel Filan, for example. It’s also led by Stuart Russell, who is probably the figure most associated with AI x-risk research amongst the AI research community, with perhaps the exception of Nick Bostrom.
If you think that I’m better at reducing x-risk than they are (e.g. because my beliefs are better), then that seems like a plausible reason to worry. I do expect that the research that CHAI does will be somewhat different now that I am not there; I have many disagreements with many people at CHAI. But I wouldn’t worry about them being distracted.
I personally donated to CHAI this year; part of my reasoning was that CHAI could use unrestricted funding for research engineering roles. (Note that I am now at DeepMind, so I was less worried about conflicts of interest, though obviously I still have many friends at CHAI.)
I wanted to note I agree with most of this section, and have adopted a similar policy. I do think there are more downsides to calling out poor research—the main one I’d note is that in general because communication is hard people (or at least I) tend to be systematically too negative about other people’s research. (See previous link for more details.)