Blizzard and DeepMind, which was acquired by Google in 2014 and is now a subsidiary of Alphabet Inc., have just announced opening up StarCraft II for AI research. DeepMind was the company that made AlphaGo, which beat Lee Sedol, a grandmaster of Go, in a best-of-five showmatch with a score of four to one. They hinted at possibly having a BlizzCon champion, some year, do a showmatch as well, which would be entertaining.
StarCraft II is different from Go in three important ways. First, any given player knows what they scout, which they apparently will constrain these AI to honor. Second, there are three possible match-ups for any choice of race, except random, which has nine. Third, it's real-time, which can be good for AI, because they're not constrained by human input limitations, but also difficult from a performance standpoint.
From Blizzard's perspective, better AI can be useful, because humans need to be challenged to learn. Novices won't be embarrassed to lose to a computer over and over, so they can have a human-like opponent to experiment with. Likewise, grandmasters will want to have someone better than them to keep advancing, especially if it allows them to keep new strategies hidden. From DeepMind's perspective, this is another step in AI research, which could be applied to science, medicine, and so forth in the coming years and decades.
Unfortunately, this is an early announcement. We don't know any more details, although they will have a Blizzcon panel on Saturday at 1pm EDT (10am PDT).
Awesome but I can see myself
Awesome but I can see myself screaming at the AI for always winning. I mean how can you bet a computer that processes large amount of data and statistics way faster than you. They better have it oblivious to the enemy economy, military strength, and positioning just like a human would be in game. Basically it has to figure all that out on its own.
Having it make decisions with
Having it make decisions with hidden information is one of the goals of the research, apparently. Also, I'm sure that, if they can make an AI that destroys grandmasters by deep learning, they can force it to make human-like mistakes if they gave it a good try.
With Alphabet(Owns Google)
With Alphabet(Owns Google) that AI may just be blinding you with ad content and causing you to lose. I can see the AI trying to strike up a conversation while you both battle it out to get at what products you may be wanting!
[If you like this product, distroy tank number 3]! Any urban warfare and those billboards will be pushing real products! [Do this survey to come back to life! Otherwise Game Over!]
Do the handful of people
Do the handful of people still playing SC2 even care about AI performance?
I always thought it was ladder/multiplayer centric. I haven’t touched the game since 2012 so if something has changed I’d love to know.
I pretty much only play
I pretty much only play Arcade because I have an awful short-term memory and forget to macro. About a year ago, they launched co-op campaigns that are pretty popular, though.
Also, while this is probably more research project than gaming product, I'm probably one of the people who would like to play against AI if it was a human-like experience. Especially for team games, it would be fun to mess around, especially with odd strategies, without griefing teammates that actually want to win or wrecking a persistent score.
I think that the purpose is
I think that the purpose is for AI research, not for the benefit of the players. Starcraft II requires the processing of a lot of unsorted data, analysing a situation with incomplete information, having many different factors impact different situations, bluffing, etc. A lot of things that aren’t trivial for AIs.
I follow the SC2 scene and
I follow the SC2 scene and play a decent amount. The DeepMind partnership for SC2 AI research is mainly for promotional and research purposes.
There really isn’t a benefit for the multiplayer scene in having AI to play against, since there is still a healthy player population. Also, the current vs AI modes like Coop benefit from the relatively simple AI that is currently used. Those modes are supposed to be relaxing, which means better AI would actually make them worse.
I disagree with your latter
I disagree with your latter paragraph. You're describing the current use cases for StarCraft II AI, and that's perfectly valid, but you're then claiming it shouldn't be used in other ways, too (assuming they get it to work).
As I've said earlier, it could be useful at the very top end, if it can beat grandmasters, as a replacement for barcoding on ladder. Also, I'd expect they would be able to scale down to various levels of AI, neutering it for all player skill levels. If they can do any or all of that, then there's several other unserved use cases that this AI could fulfill.
But yes, I expect it's mostly a research project.