Wasn't YC always explicit (at least if you read between the lines) about looking for founders with good pedigree regardless of having a feasible product?
Yes and that makes sense as an investment thesis. But what happens when you attract a bunch of ambitious people with great pedigree and then either steer them or select for jumping onto a hype bubble at it's peak with an undifferentiated cobbled together offering?
That's what concerns most about this, is how somehow it looks like they've optimized for a bunch of "follower" founders.
But I really wonder how many of these founders can convincingly explain in 5 minutes how gradient descent or transformers work. I know that's the first type of question I would ask any founder of an AI company before investing.
The typical thinking among businesspeople is that you can always "hire a nerd" for those things. The early YC was a departure from that with PG praising technical prowess in his essays so much. It would seem this mindset has much less sway now.
The combination of gradient descent and transformers is a weird choice to ask founders about. One is an almost trivial algorithm, whereas the other is a cutting-edge research result.
That's what concerns most about this, is how somehow it looks like they've optimized for a bunch of "follower" founders.