Hacker News Reader

Comment by Berobero | Hacker News Reader

Counterpoint: this article (https://pdfs.semanticscholar.org/5a12/80f783e4ce6ba31b821f4d...)...

TL;DR: Experts have a poor track record for predicting AI development, they aren't particularly qualitatively different in their predictions from non-experts, and "15-25 years" has been a staple prediction for decades now.

View on HN

Counter-counterpoint - Kurzweil's project Moore's law technique predicted computers being chess champ within a year of the event and has predicted passing the Turing test for 2029 for ages now, which I guess if roughly equivalent to general intelligence.

There is a certain irony in saying we should believe in Kurzweil just because his n=1 predictions came true.

(Statisticians often deride machine learning people for having a poor understanding of statistics).

There's more to the Moore's law argument than the one prediction from Kurzweil. For example there's Hans Moravec's essay from 97 predicting "the required hardware will be available in cheap machines in the 2020s" and I remember writing about the stuff for my uni entrance exam in 1981 before I'd heard of Kurzweil or Moravec. The basic idea is that when brain equivalent hardware is available cheaply so hackers can hack on it, the software will follow not long after if not before. Given the recent progress and the sheer number of talented people flooding into AI research that seems quite likely to me.

Reply on news.ycombinator.com