OpenAI says it’s planning to create a team to help manage the risks that could come from a superintelligent AI system that it expects to arrive within the decade.