I think that it would be great if MIRI had direct competitors. We would be better able to see what is essential to AI ethics and what is a local parochialism.
I am challenged to think of a way that my Newly Minted MIRI Competitor could differentiate itself from MIRI in a way that is both:
a) optimal, and
b) non-divergent from MIRI’s goals in such a way as to be functionally different.
I could certainly fund Evilbot Angry AI Development Labs, and we might see a difference in focus away from GAI frameworks and more towards how-to-kill-all-the-humans-as-effectively-as-possible research, but that doesn’t let me weed out what ‘local parochialism’ is.
I suggest that the more Musk influences OpenAI’s agenda, the further it moves away from core competition with MIRI.
A counterexample might be if a series of AI researchers in China announced a formation a clone of MIRI but based out of Shanghai—a more clear-cut intelligence race than what we’ve currently go, which is an increasing number of institutions all starting down roadmaps that share initial common ground but have divergent ideal end states.
I think that it would be great if MIRI had direct competitors. We would be better able to see what is essential to AI ethics and what is a local parochialism.
I am challenged to think of a way that my Newly Minted MIRI Competitor could differentiate itself from MIRI in a way that is both:
a) optimal, and
b) non-divergent from MIRI’s goals in such a way as to be functionally different.
I could certainly fund Evilbot Angry AI Development Labs, and we might see a difference in focus away from GAI frameworks and more towards how-to-kill-all-the-humans-as-effectively-as-possible research, but that doesn’t let me weed out what ‘local parochialism’ is.
FHI is a MIRI competitor. OpenAI is now also a MIRI competitor.
I suggest that the more Musk influences OpenAI’s agenda, the further it moves away from core competition with MIRI.
A counterexample might be if a series of AI researchers in China announced a formation a clone of MIRI but based out of Shanghai—a more clear-cut intelligence race than what we’ve currently go, which is an increasing number of institutions all starting down roadmaps that share initial common ground but have divergent ideal end states.
No one knows what is optimal, if more people attempted to find it, it would increase probability that at least one of them would succeed.