> Obviously it is better to be globally optimal than locally optimal.
the implication of having machines dictate our behavior [WRT environment, each other] to optimize the survival of humans far into the future seems a bit closer to reality.
I don't want to downplay the amazing breakthrough that AlphaGo represents, but I think it's a bit melodramatic to extrapolate from a program winning a board game to AI's impact on humanity's survival.
it's a board game of military strategy. while it is severly limited by its rules of play relative to real life, it is not inconievable that at some point it can dictate life and death as a capable military advisor.
i think it is inevitable that we will utilize AI to make better predictions about our own future than weather forecasters. not in our lifetimes, of course.
the program that beat Kasparov did so using a brute-force search (full?) of a very limited search space. the reason that AI is different is because it doesnt need to (and cannot do so because the space is simply too large.)
and 20 years ago is still well "within our lifetimes"
The program that beat Kasparov avoided a full breadth first search because it was fairly easy to decide if something was a bad move. So, it could generally pick from a fairly small number of moves when branching. What makes GO so much harder than chess is how hard it is to evaluate the board and decide if a move is good or bad.
First order branching factor is less important than you might think.
You could setup a game of Nim: https://en.wikipedia.org/wiki/Nim with much higher branching factor vs Go, but because you can trivially separate a winning moves vs. losing moves those branches become irrelevant in Nim. AKA, if there are 2,000 winning moves then they are all equivalent.
how do you know that AI is not already advising the military on certain tactics? even an isolated system that analyzes data gathered by intelligence analysts, weather, radar, etc. and can suggest an a flight path and timeline for insertion or airstrike (especially in context with other operations in-theater) would basically be that, wouldn't it? i mean, i doubt it's guys on the DoD version of google maps plotting routes by hand.
i have a hard time believing something like that isn't already in use to some extent, at least for special operations forces. how long until they just hook that into the drone fleet? what if they already have? autopilot on a modern fighter/bomber is basically the same thing, except there's a human in the loop, right?
that would also go along nicely with the whole matrix / skynet plot that seems playing out in real life right now...
Humans are better at thinking globally than they are at acting globally. It's not clear that a "wise" AI would be helpful to us. It would tell us to do things that we already know we ought to be doing, but that we don't do for local, near-term-selfish reasons.
this is true, but it mostly comes down to 1. don't obliterate each other, 2. don't destroy your dependencies, 3. don't out-consume your dependencies (overpopulate, etc).
however, point 2 is only understood from a very high level by humans. we know the dependency chain goes deep, but we don't know exactly which branches can be ignored without effect and which are critical.
our default is to protect all the things "just in case"
"Optimization" presumes the existence of concrete criteria by which an outcome may be judged, and often a timeframe within which to judge it (e.g. if I want to maximize my wealth when I retire I should fully fund my 401(k), but if I want to maximize my wealth next year then I shouldn't fund it at all). A board game has unambiguous criteria for victory; human civilization, not so much.
the implication of having machines dictate our behavior [WRT environment, each other] to optimize the survival of humans far into the future seems a bit closer to reality.