Teaching A.I. Systems to Behave Themselves

As philosophers and pundits worry that artificial intelligence will one day harm the world, some researchers are working on ways to lower the risks.