5

17 comments

[–] smallpond 2 points (+2|-0)

I was wondering if this extreme obscure skill is transferable to something I'd more readily recognize as useful.

Being a drone pilot crossed my mind, but then AI has basically bested us at that. Has AI also blown humanity away at Mario Kart 64?

[–] PMYA [OP] 2 points (+2|-0)

I think it is transferrable. The level of patience it takes to get to the point where you are playing any game at this level is something that is not easy to achieve. In some ways I would compare it to learning a martial art.

AI does not play Mario Kart. You can teach AI to play chess or something, but the amount of variables in MK64 is probably beyond an AI at this point. There is something called a TAS, or Tool-Assisted Speedrun, where the game is "played" perfectly by a predetermined series of inputs created by a human, that's the closest speedrunning comes to AI.

[–] smallpond 2 points (+2|-0) Edited

Well, I'm not questioning the skill/patience/dedication involved (which are transferrable) just the transferability of the actual skills developed.

AI is now beating fighter pilots at combat simulations. If MK64 isn't already within its reach, I think it soon will be.

Edit: Too much 'skill' above, but you should get my meaning.

[–] PMYA [OP] 2 points (+2|-0)

At the risk of sounding stupid, I'm still going to say that in this scenario, MK64 is more complex than flying a fighter plane. First off, it's not exactly clear what kind of decision making is going on with this AI. Is it choosing from some kind of library of manoeuvres that it did not come up with itself, or is every input being calculated by the AI? I would be willing to bet it is the former, as the level of complexity needed to do the latter just does not exist.

The margin for error in a speedrunning setting, particularly in a 3D game, is a lot slimmer than beating a human pilot. If an angle is not calculated properly when a plane is being flown, it is not necessarily a mistake if the only objective is to beat the other pilot, there is adjustment time. Every miscalculation in MK64 is an error that can not be recovered, as it is lost time. Small errors can also lead to insanely large timelosses if they happen in the wrong places. There are also a lot of broader things to think about in an MK64 race too, such as where and when to use certain items optimally. I do not think an AI could calculate all of this in real time better than a human, for the time being.

How long would it take for an AI to teach itself how to do this?