There is a blind spot in AI research

Chicago police use algorithmic systems to predict which people are most likely to be involved in a shooting, but they have proved largely ineffective. This week, the White House published its report on the future of artificial intelligence (AI) -- a product of four workshops held between May and July 2016 in Seattle, Pittsburgh, Washington DC and New York City (see go.nature.com/2dx8rv6). During these events (which we helped to organize), many of the world's leading thinkers from diverse fields discussed how AI will change the way we live. Dozens of presentations revealed the promise of using progress in machine learning and other AI techniques to perform a range of complex tasks in every day life. These ranged from the identification of skin alterations that are indicative of early-stage cancer to the reduction of energy costs for data centres. The workshops also highlighted a major blind spot in thinking about AI.