Notes on "How do we become confident in the safety of a machine learning system?" 2023-10-26T03:13:56.024Z

Quick Thoughts on Language Models 2023-07-18T20:38:57.866Z

~100 Interesting Questions 2023-03-30T13:57:28.591Z

A Thorough Introduction to Abstraction 2023-01-13T00:30:25.964Z

Content and Takeaways from SERI MATS Training Program with John Wentworth 2022-12-24T04:17:21.094Z

Follow along with Columbia EA's Advanced AI Safety Fellowship! 2022-07-02T17:45:47.120Z

Comment by RohanS on ~100 Interesting Questions · 2023-03-31T21:50:48.846Z · LW · GW

Lots of interesting thoughts, thanks for sharing!

You seem to have an unconventional view about death informed by your metaphysics (suggested by your responses to 56, 89, and 96), but I don’t fully see what it is. Can you elaborate?

Comment by RohanS on ~100 Interesting Questions · 2023-03-30T18:13:01.876Z · LW · GW

Basic idea of 85 is that we generally agree there have been moral catastrophes in the past, such as widespread slavery. Are there ongoing moral catastrophes? I think factory farming is a pretty obvious one. There's a philosophy paper called "The Possibility of an Ongoing Moral Catastrophe" that gives more context.

Comment by RohanS on Information Loss --> Basin flatness · 2022-11-15T01:27:06.117Z · LW · GW

How is there more than one solution manifold? If a solution manifold is a behavior manifold which corresponds to a global minimum train loss, and we're looking at an overparameterized regime, then isn't there only one solution manifold, which corresponds to achieving zero train loss?