Stop Deploying General-Purpose AI Models For Everything

Ensuring safe A.I. is another reason developers should stop deploying general-purpose models for everything. To date, the industry has been unable to guarantee that generative A.I. systems will stick to their safety instructions. Studies have documented instances in which generative A.I. deceives its human operators, tries to use blackmail if its self-preservation is threatened and responds in a way that could lead to murder. More specialized systems like AlphaFold and Waymo’s driving systems won’t misbehave that way because their operating parameters are much narrower. - Gary Marcus writing in the New York Times