Thread

1. elzbardico · 1790719910 · focus · HN ↗

   A benign but depressing interpretation is that even competent and respected academic researchers are not immune to AI-addiction and episodes of AI-psychosis cause by binging on AI.

   The dopamine hit of having even their most speculative ideas expanded in a convincing paper format is self-reinforcing and soon the affected individual starts believing that the output is perfect and needs no review, because reviewing it would delay the next stimulation cycle.

   1. Barrin92 · 1790722869 · focus · HN ↗

      I wouldn't consider that benign but I agree with the characterization, I've come to see the whole AI psychosis thing as a version of sports betting, gacha videogame mechanics or prediction markets. In the US slightly faster than anywhere else pulling the slot machine lever one more time seems to be increasingly become the sole principle of any activity.

      Correctness, formal verification, using your own cognitive resources, deterministic results are in many ways actively counter productive to the kind of effect people want to get out of this

      1. elzbardico · 1790724408 · focus · HN ↗

         Oh sure, just to clarify, I said benign in the sense of not attributing malicy to the researcher actions. But in fact, the AI psychosis scenario is a frankly scary possibility.

Unofficial Hacker News client; not affiliated with Y Combinator.