Neo CEO Ali Partovi discusses his proposal for assigning responsibilityGovernance defines responsibilities, decision rights and controls for how a system or organization is operated. when an AI system independently takes an unlawful action. He argues that companies should face penalties associated with intentional conduct to create stronger incentives for prevention.
In conversation with MTS host Theo Jaffee, Ali Partovi distinguishes an action directed by a human from one selected by an agent. Asked about difficult cases, he says courts would need to examine the evidence and suggests human authorizationHuman-in-the-loop command approval requires a person to review and authorize a consequential AI action before it executes. could create a record of responsibility.
Ali Partovi presents the proposal as a potential contribution to AI alignmentAlignment aims to make an AI system's behaviour consistent with intended human goals, values and constraints. while acknowledging uncertainty about whether it would be sufficient. His claims about law and future outcomes are presented as a policy proposal, not a statement of existing law.
Watch on YouTube
