Pacing the AI Frontier: the risk of artificial unintelligence

The most immediate risks stem not from artificial superintelligence but from artificial unintelligence