Evidence receipt / belief
Published · transcript-backedSholto Douglas: belief
28 Mar 2024 Dwarkesh Podcast Sholto Douglas & Trenton Bricken — How LLMs actually think
“” There's also someone who works on Anthropic's performance team now, Simon Boehm, who has written in my mind the reference for optimizing a CUDA map model on a GPU.”
Source trail
Everything needed to verify it.
- Speaker
- Sholto Douglas
- Attribution
- Verified speaker
- Claim type
- belief
- Recorded
- 28 Mar 2024
- Publisher
- Dwarkesh Podcast
Transcript context
…One of the other things I want to emphasize for a potential slice of the audience is that there's this sense that the world is legible and efficient, that you just go to jobs.google.com or jobs.whatevercompany.com and you apply and there's the steps and they will evaluate you efficiently. Not only from your stories, but it just seems like often that's not the way it happens. In fact, it's good for the world that that's not often how it happens. It is important to look at, “were they able to write an interesting technical blog post about their research or are they making interesting contributions.” I want you to riff on this for the people who are assuming that the other end of the job board is super legible and mechanical. This is not how it works and in fact, people are looking for the different kind of person who's agentic and putting stuff out there. I think specifically what people are looking for are two things. One is agency and putting yourself out there. The second is the ability to do something at a world-class level. There are two examples that I always like to point to here. Andy Jones from Anthropic did an amazing paper on scaling laws as applied to board games. It didn't require much resources. It demonstrated incredible engineering skill and incredible understanding of the most topical problem of the time. He didn't come from a typical academic background or whatever. As I understand it, basically as soon as he came out with that paper, both Anthropic and OpenAI were like, “we would desperately like to hire you. ” There's also someone who works on Anthropic's performance team now, Simon Boehm, who has written in my mind the reference for optimizing a CUDA map model on a GPU. It demonstrates an example of taking some prompt effectively and producing the world-class reference example for it, in something that wasn't particularly well done so far. I think that’s an incredible demonstration of ability and agency and in my mind would be an immediate, “we would please love to interview/hire you.” The only thing I can add here is I still had to go through the whole hiring process and all the standard interviews and this sort of thing.…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.