Demand “Explainability” When Contracting for AI

The promise of AI systems shouldn’t be ignored, and neither should the limitations and risks that come with it. A post from law firm Pillsbury outlines potential risks, in particular from AI that turns out to be unreliable or discriminatory, and suggests what companies considering contracting with an AI provider can do to steer clear of them.

The writers recommend both discussion and building protection into the contract. Among the elements that might be addressed are the specs of the AI system being considered and how they address potential discrimination and fact-checking issues, and indemnification obligations in the event of discrimination or other claims.