Generative AI can produce impressive work, confident errors, unexpected ideas, and unsettling substitutions in the same session. Ethan Mollick argues that the useful response is neither blind adoption nor distant refusal, but active experimentation with clear human responsibility.
New York Times Bestseller.
Mollick writes from inside a fast-moving experiment: generative AI is capable enough to alter work and education, yet unreliable enough that polished output cannot be accepted at face value. He organizes that tension through the jagged frontier of capability and four working rules, then examines AI as a person-like collaborator, creative partner, colleague, tutor, and coach. The argument repeatedly returns to what humans should retain: judgment, accountability, expertise, and a reason to care about the work.
The guide shows where broad experimentation reveals useful capabilities and where privacy, expertise, or consequence requires tighter boundaries.
AI competence varies unpredictably across tasks. Readers learn to test the actual task rather than infer reliability from a single fluent answer.
Delegation changes work, education, and creative identity; the chapter map keeps efficiency questions connected to accountability, meaning, and who makes the final decision.
Readers can compare tasks where AI expands iteration or access with tasks where hallucination, bias, privacy, deskilling, or false confidence becomes decisive. The chapters also help separate observations made with current systems from longer-range forecasts, an essential distinction in a field where product capabilities and institutional policies change rapidly.
Educators, managers, creators, and knowledge workers who already encounter generative AI will gain the most. The value lies less in fixed prompts than in a disciplined way to test tools while keeping a human in the loop.
Edition notice: This is an independent chapter guide and summary, not the original book. It is designed for review and reading support and is best used alongside the full work.
Miva: the next reading experience beyond audiobooks
Listen like an audiobook, then ask anything, anytime.
Start Miva's narrationMiva does not just narrate. It keeps explaining based on what you ask.