3 ms·
Even as a massive sceptic of Cyberpunk AI (and for that matter, the efficacy of preventing it by setting up charities to do AI research), I think people are per
by notahacker 8y ago
Even as a massive sceptic of Cyberpunk AI (and for that matter, the efficacy of preventing it by setting up charities to do AI research), I think people are perfectly entitled to spend their money on it if it's what they're interested in. I also think it's plausible some such programmes might create extremely commercially valuable byproducts even though they don't lead anywhere useful directly, much like the Apollo missions.
Trouble is, if you invest enormous amounts of time and energy in arguing that a general belief stuff might work isn't good enough for philanthropy and even writing "why we don't recommend" articles about specific charities dispensing aid that don't deprive enough people of it to have a control group for RCTs, you deserve every criticism you get when you then funnel foundation money to very well-funded AI startups with unfalsifiable solutions to a purely hypothetical problem, even if you didn't have close personal relationships with people working for them. Even though I consider some EA analysis to be good and well reasoned (and they're certainly not the only people making evidence based bearish arguments about, say, microfinance) it's a little difficult to see how AI research could possibly pass a decision-making heuristic so supposedly rigorous that it writes off sanitation as an area to invest in because the estimations of diarrhea reduction aren't blinded. Ultimately, EAs have their cognitive biases just like anyone else, and I'm more sympathetic than they are to aid organizations' view that conducting RCTs is difficult and expensive and they don't want people suffering in control groups to quantify how well common sense healthcare interventions work, and disagree that such organizations should be held to higher standards of analytical rigour than AI researchers hoping to be Sarah Connor.