Advances in artificial intelligence (AI) are a double-edged sword. On theone hand, they may increase economic growth as AI augments our ability toinnovate. On the other hand, many experts worry that these advances entailexistential risk: creating a superintelligence misaligned with human valuescould lead to catastrophic outcomes, even possibly human extinction. Thispaper considers the optimal use of AI technology in the presence of theseopportunities and risks. Under what conditions should we continue the rapidprogress of AI, and under what conditions should we stop?
