What's this about?
Dan Luu checked how the most widely-cited AI skeptic's predictions have actually panned out. He discloses his own position first, and it is unusually well-documented: in 2022 he audited futurist predictions including Kurzweil's and found them generally wrong on both results and reasoning; in 2015 he wrote that people were underestimating AI's ability to displace human jobs. His summary of his own view: "if something is currently happening, the people who are saying that it's impossible that it will ever happen are probably wrong."
Why does the disclosure section matter so much?
Because the standard rebuttal to any AI-skeptic critique is that the critic is financially interested. Luu pre-empts it in detail — no lab employment, no supplier, index funds only, seed investments underweight on AI, and a failed AI lab interview he describes at length. Whether or not you find the analysis convincing, the methodology of stating your priors and your incentives before scoring someone else's predictions is the part worth copying.
What's the community saying?
- The most common defection story: people who followed Zitron because he seemed to do the research, until they noticed the numbers only got engaged when they served the argument.
- The structural critique: he is now part of the professional social-media class and makes money through engagement, which is a reason to weight him accordingly.
- The most balanced position, and the most upvoted: he is right about OpenAI being over its skis, right that the numbers are wildly optimistic, and right about the circular financing — but he hates AI, so he cannot acknowledge its usefulness.
- The refinement that follows: his *thesis* is roughly correct and his *predictions* are irrelevant, because nobody can predict a field this dynamic.
- A long sub-argument about whether "a bubble will pop" is a reasoned position or a religious one, with the gambling framing — the odds a bubble never pops again are not good — winning on points.