Researchers propose theory for misbehaving AI agents

The makers of A.I. want to build systems that follow human direction, without resisting or deceiving their operators. Open article →