Posts
Comments
Anyone who wants to work in AI safety needs a good answer to the application question “tell me about a technical project you’re proud of, preferably related to AI safety.” It may sound obvious, but as a student who hasn’t worked on projects as much as I should over the past few years, I wish somebody had told me this earlier. Having an independent project, even if it’s just proof that you can use RL, looks quite important for getting past many first stage application processes.
This looks like an interesting subject/interaction that I would like to learn more about, however reading the unedited output of an extremely long LLM conversation is not enlightening. I think I would really enjoy reading another post by the author analyzing this interaction and its implications, but I am almost sure that wading through close to an hour of raw AI slop is not the right medium to express this idea. I agree with below comments about this dialog being closer to experimental data than something worth posting on its own.