Shawn Wang: belief

Transcript context

…Yeah, I think that's my whole, I did this post, it's pinned on my Twitter, so you can find it easily, but about skill floor and skill ceiling in jobs. And I think the skill floor more and more, I think 2025 will be the first year where the AI sets the skill floor. Overall, you know, I don't think that has been true in the past, but yeah, I think now really, like, you know, if Devon works, if all these customer support agents are working. So now to be a customer support person, you need to be better than an agent because the economics just don't work. I think the same is going to happen to in software engineering, which I think the skill floor is very low. You know, like there's a lot of people doing software engineering that are really not that good. So I'm curious to see it. And the next year of the recap, what other jobs are going to have that change? Yeah. Every NeurIPS that I go, I have some chats with researchers and I'll just highlight the best prediction from that group. And then we'll move on [01:42:00] to end of year recap in terms of, we'll just go down the list of top five podcasts and then we'll end it. So the best prediction was that there will be a foreign spy caught at one of the major labs. So this is part of the consciousness already that, uh, you know, like, you know, whenever you see someone who is like too attractive in a San Francisco party, where it's like the ratio is like 100 guys to one girl, and like suddenly the girl is like super interested in you, like, you know, it may not be your looks. Um, so, um, There's a lot of like state level secrets that are kept in these labs and not that much security. I think if anything, the situational awareness essay did to raise awareness of it, I think it was directionally correct, even if not precisely correct. We should start caring a lot about this. OpenAI has hired a CISO this year. And I think like the security space in general. Oh, I remember what I was going to say about Apple Foundation Model before we cut for a break. They announced Apple Secure Cloud, Cloud Compute. And I think, um, We are also interested in investing in areas [01:43:00] that are basically secure cloud LLM inference for everybody. I think like what we have today is not secure enough because it's like normal security when like this is literally a state level interest. Agreed. Top episodes? Yeah. So I'm just going through the sub stack. Number one, the David one. That's the most popular 2024. Why Google failed to make GPT 3?…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.