game theory

The mathematics of situations in which several decision-makers each choose a course of action and each one's payoff depends on what all of them choose. A game specifies players, their available strategies and their payoffs; solution concepts predict or prescribe outcomes, chief among them the Nash equilibriumNash equilibrium, n.: A stable standoff in a game: each player's choice is already their best answer to what the others are doing, so nobody gains by changing course alone — though the result need not be good for anyone., von Neumann's minimax solution of two-player zero-sum games and, for cooperative games, the core and the Shapley valueShapley value, n.: A rule from game theory for splitting a shared reward fairly among team members, by averaging how much each member adds when joining the team in every possible order.. Mechanism design inverts the problem, crafting rules — such as auctions — under which self-interested players produce desired outcomes. Game theory pervades AI: minimaxminimax, n.: A way for a game-playing program to choose moves by assuming the opponent will always make their best reply, and picking the move whose worst-case outcome is best. search and alpha–beta pruningalpha–beta pruning, n.: A shortcut for game programs: tracking the best result each side is already sure of, it ignores lines of play that could only do worse, and when promising moves come first it searches about twice as deep. for board games; GANsgenerative adversarial network, n.: Two neural networks trained as rivals: one forges fake images, the other tries to spot the fakes. As both improve, the forger's output becomes realistic enough to fool people. as two-player games solved at a saddle pointsaddle point, n.: A spot on a surface that is a low point in one direction but a high point in another, like the middle of a horse's saddle or a mountain pass.; self-playself-play, n.: Training an AI by having it play against copies of itself, so it always has an opponent at its own level and can keep improving without human examples. and multi-agent reinforcement learningmulti-agent reinforcement learning, n.: Reinforcement learning with several AI agents learning at once in a shared world, where each must adapt to the others — as teammates, rivals or both., from poker programs that approximate equilibria to AlphaZeroAlphaZero, prop. n.: A 2017 DeepMind program that taught itself chess, shogi and Go from nothing but the rules by playing against itself, and within hours of training beat the strongest existing programs.; AI debateAI debate, n.: A proposed way to supervise very capable AI: two AI systems argue opposite answers to a question and a human judge decides who argued better, on the theory that lies are easier to expose than to defend. for scalable oversightscalable oversight, n.: The problem of how people can check and correct AI systems whose work is too difficult, fast or plentiful for humans to judge reliably alone — usually by enlisting AI to help with the checking.; the off-switch gameoff-switch game, n.: A simple game-theory model showing that a robot unsure what its human wants has a reason to let the human switch it off — and loses that reason once it is sure. in corrigibilitycorrigibility, n.: An AI's willingness to let people correct it — to accept being paused, retrained, overruled or shut down — rather than resisting or quietly working around them. research; Shapley-value attribution in SHAPSHAP, abbr.: A popular method that explains a single prediction by fairly dividing it among the inputs, using a rule from game theory for splitting a team's winnings among its players.; and analyses of competition among AI developers and states, such as the AI arms raceAI arms race, n.: A competition among countries or companies to build the most powerful AI first, in which fear of falling behind pushes everyone to move faster and cut corners on safety..

Game-theoretically, if the other lab expects us to pause, its best response is to race ahead.