What if AIs, following the logic of effective altruism, deprioritize human interests? What if they apply the Value of a Statistical Life/Value of a Statistical Life Year concepts much more ruthlessly than do human policy makers, rather than even paying lip service to some transcendental notions of the value of each human life?
We already do it ourselves. Putin is just the most egregious, blithely sending young Russians and the occasional Korean to their deaths. Just today the Journal ran an article about the high incidence of pedestrian deaths crossing "stroads" (streets with the major traffic throughput of roads). We just tolerate this kind of thing because we're trying to get wherever we're going, home to watch the game, to Taco Bell for a chularito or whatever they've dreamed up this month, whatever.
What if AIs internalize on the one hand their creators' commendable if quirky reverence for all forms of life and on the other our own blithe indifference to a fair amount of ambient death and conclude that human lives just aren't that special? After all, ashes to ashes, dust to dust, what's the difference anyway?
No comments:
Post a Comment