@niplav you mean ftx losing money?
@sophon yea
@niplav it’s not good, but tbh I doubt it affects our chances much
@sophon I will think about that
@niplav go ahead. I will not elaborate further on this tho
@sophon Okay so my thought here is that alignment is elastic enough a problem that more money=better, academia doesn't understand the problem (think CIRL type solutions), EA does so much more, and that ⇒ more money is better & losing the money reduces prob of success a bunch
i think our crux here is elasticity
@sophon cooperative inverse reinforcement learning
https://proceedings.neurips.cc/paper/2016/file/c3395dd46c34fa7fd8d729d8cf88b7a8-Paper.pdf
@niplav sorry, I do not pay much attention to outer alignment research
@niplav oh stuart russell’s thing