And for whatever reason a lot of people in startup/tech seem to have a huge Dunning-Kruger effect blind spot where they believe knowing a lot about one thing makes them an expert in everything.
This used to just be funny, but when it started to intersect with politics it began to actively contribute to destroying society. It isn't funny anymore.
(I don't think Karpathy's job data here is destroying society, this is a more generalized observation).
This is the equivalence of telling a Designer that can't create infographics on anything but principled design subjects -- or else they're out of line. Any research or data they might use isn't relevant because they're not exerts? lol?
It is a website that visualizes the output of an LLM prompt and passes it off as data. Big difference between the two.
Its especially(!) very common for people who made an exit and are now "wealthy" - sure they can afford to have an oppinion on everything, but very often they are just talking bullshit, thinking: "hey, I made it in field X, so why do not try field Y".
Esp the "MBA crowd" is famous for this: For whatever reason they think they are more intelligent than ana engineer who filed a patent, e.g. (while most of the MBA bobos would fail just in acquiring all documents required for this)
Other example: If you wrote once a book and it got traction, even if you are not a proven expert you will be invited to television shows etc. (and MORE than the people who are real experts with proven track record)