Ethan Mollick Shares Mental Model for AI Delegation

Ethan Mollick posted guidance on working with Fable and Astra class models. He advises treating them as a capable outside team rather than an intern. Users should state desired results explicitly, set the amount of leeway granted, identify what needs testing, note when the model should ask for help, and define what good output looks like. Mollick is an associate professor at Wharton who studies AI and innovation. The post offers practical steps for directing these models on complex work.