How DeepMind plans to stop AI from behaving badly

Researchers at the Alphabet subsidiary DeepMind have spelled out how they will ensure that AI is developed safely. The guidelines aim to make certain that powerful systems capable of learning and figuring out their own solutions to problems don't start to behave in unexpected and unwanted ways. The big issues: The researchers say the key challenges are specifying the intended behavior of a system in a way that avoids unwanted consequences; making it robust even in the face of unpredictability; and providing assurances, or ways to override behavior if necessary. Erratic behavior: This is a growing area of academic research. There are plenty of often amusing examples of machine-learning systems that have started behaving oddly.