On: OpenAI Sold its Soul for $1B

[Kalanos]

Anything they do other than writing a morally-incentivized governance system for reinforcement models (aka pushing an API for a trillion param NLP model) is just fake.

[xkapastel → Kalanos]

Could you expand on this? I can't really parse it; the first thing that came to mind was a DAO where token holders could vote on...something related to parameters and fine tuning, but idk what.

[Kalanos → xkapastel]

safe reinforcement learning that won't turn the universe into a paperclip factory