He attributes public AI distrust to past cryptocurrency hype.
Positive users back Scott Alexander's call for ruthless truth in AI by praising experimenters as heroes, while negative users worry it boosts charlatans and funds blind alleys.
No Digg Deeper questions have been answered for this story yet.
@slatestarcodex @tszzl @cremieuxrecueil I just don't get why this is important at all. If it doesn't work they will lose money and it won't be adopted.
People who try new things and fail are heroes. We should cheer them on. If it's a fraud ofc that's a different story but zero reason to think that atm. Let them cook
@slatestarcodex @tszzl @cremieuxrecueil Boosting every technology means going down a lot of blind alleys and also funding a great many charlatans
@slatestarcodex @tszzl @cremieuxrecueil If there's an actual story of fraud here I'd love to hear it. Otherwise you're just throwing rocks from the sidelines
Correct me if I'm wrong, but I took roon's proposal to be "we should call out obvious BS and ground predictions in reality, but we should be like 30% less harsh and cynical than we otherwise would be in order to encourage experimentalism, nonconformity, and innovation".
Like, rather than being a universal booster for everything, I'm imagining that Scott and roon agree about criticizing the very worst ideas and praising the very best ideas, but for meh ideas like MidJourney Scott wants all the savvy people to say "meh" and roon wants all the savvy people to say "oooh, neat! not sure this'll work out, but worth exploring" (or keep their negativity to themselves), so that savvy people don't harshly socially punish experimentation, but they do stay informative (especially in extreme cases).
If that's true, I wonder if a solution might be to valorize unlikely experiments more, so we can loudly call things 'unlikely to pan out' without this always seeming inherently shameful? (E.g., maybe we should call more things "moonshots".)