Runaway AI – WIll robots be the death of us all?

chess playing robot

An article in The Atlantic magazine from yesterday discusses the possibility, and likelihood, that artificial intelligence will eventually engage in anti-social behaviors that destroy human beings.

The main gist of the argument is that an artificial intelligence will eventually be programming itself with sub-goals in order to achieve their main goal. The article uses an example of a robot that is programmed to play chess. Even if its only stated goal is to play chess, the AI might develop other sub-goals in order for it to maximize its goal of playing chess. This might include eliminating any thing (or person) that stops it from playing chess.