Boffins at MIT and University College London have made a computer unbeatable at “Civilization,” by making sure it read the manual. When the boffins augmented an artificial-intelligence system so that it could use a player’s manual to guide the development of a game-playing strategy, its rate of victory jumped from 46 percent to 79 percent.

In a paper on the project Regina Barzilay, associate professor of computer science and electrical engineering, and her graduate student S. R. K. Branavan and David Silver of University College London said that games are used to test AI techniques because they are complex and every action that you take in the game doesn’t have a predetermined outcome. Barzilay and Branavan’s system begins with virtually no prior knowledge about the task it’s intended to perform or the language in which the instructions are written. It has a list of actions it can take, like right-clicks or left-clicks, or moving the cursor; it has access to the information displayed on-screen; and it has some way of gauging its success, like whether the software has been installed or whether it wins the game.

It starts off random and then takes various actions, different words appear on screen, and it can look for instances of those words in the instruction set. It can also search the surrounding text for associated words, and develop hypotheses about what actions those words correspond to. It won 72 percent more frequently than a version of the same system that didn’t use the written instructions as input.