The Future of AI: Social AI
I’ve been talking a lot about “social AI” recently as a way to differentiate what we have been developing from “typical” or traditional AI.
The easy way to say this is “we don’t do pathfinding.” Which isn’t entirely true (we have a simple but effective pathfinding mechanism), but it shows where our focus is(n’t). Agents need to move around a world, sure; and showing crowds of agents walking purposefully about makes for a great visual demo. But to be interesting — or even more, meaningful — they need to do a lot more than that.
Another way we describe what we do is that we’re doing “heart and head” AI as opposed to “spinal” AI. I believe that to be meaningful to players, or to human users in general, AIs will have to have robust personalities, bona fide emotions (and the ability to understand your emotions), memories, relationships, the ability to learn from their experience or from what others tell them, and of course the ability to (in some form) talk about what they’re wanting, thinking, and doing. That’s a long list, but we’ve tackled most of these and are well on our way with the others.
‘Relationships’ is somewhere in the middle of that long list, and is a key aspect that shouldn’t be overlooked. AIs need to be situated in a “social context” (the next step beyond a “social network”) just as humans are. This is true in both game and non-game applications. I get very excited whenever I think of the potential for socially plausible AIs that have relationships with those around them participating in social networks/contexts. Whether this is Facebook or in an MMO, this adds hugely to the broad appeal, potential utility, and as-yet unknown forms of interaction. I can’t wait to see where this takes us.
All I know is, the future of AI isn’t better pathfinding or strafing. The future of AI is social — it’s us.