Share this comment
Just wanted to mention that if anyone liked my submissions (3rd prize, An Overview of “Obvious” Approaches to Training Wise AI Advisors - aiimpacts.org/an-overvi…, Some Preliminary Notes on the Promise of a Wisdom Explosion - aiimpacts.org/some-prel…),
I'll be running project related to this work for AI Safety Camp (Description here: docs.google.com/documen…).
Just wanted to mention that if anyone liked my submissions (3rd prize, An Overview of “Obvious” Approaches to Training Wise AI Advisors - https://aiimpacts.org/an-overview-of-obvious-approaches-to-training-wise-ai-advisors/, Some Preliminary Notes on the Promise of a Wisdom Explosion - https://aiimpacts.org/some-preliminary-notes-on-the-promise-of-a-wisdom-explosion/),
I'll be running project related to this work for AI Safety Camp (Description here: https://docs.google.com/document/d/1kJn9F_G9ezeoOrjhc4x06iv7xMuygqWWkty-ztBL6W0/edit?tab=t.0#heading=h.b8lhi4yltarg).
Just wanted to add one acknowledgement: Thanks to Anton Kalabukhov who pointed out that my previous analysis of the principled approach didn't account for the possibility of building a seed AI.