Mod note, in the spirit of our experiment in more involved moderation:
This seems a bit more grounded than many first posts about AI on LessWrong, but my guess is this doesn't meet the quality bar. See the advice in my AI Takes section of my linked comment.
Today I remembered reading about mutations in animals which allow them to become increasingly domesticated by humans. Also syndromes in humans leading to more friendly and trusting behavior
If all of these types of friendly mutations were sought out and the architectural patterns that they correspond to were mapped out.
There's a chance we could have a shot at building friendliness into an AGI system even not knowing exactly why the set up is causing it.
After reading more and more things on this site I keep cycling through stages of terror, hope, and acceptance again and again. My notions of capabilities were incredibly naive and I'm sure they still are.
It's just a shot in the dark. But, I'm just going to keep throwing out anything I can think of while I can