Takeaways
- Invite a language model to interview you by requesting it to ask successive questions until it can propose two obvious and two non-obvious AI applications for your workflows.
- Treat generative AI as a teammate: coach its output with feedback and mentorship instead of dismissing mediocre results as failures.
- Use AI to role-play and prepare for challenging conversations by having the model adopt a counterpart’s perspective and provide targeted feedback.
- Apply AI to tasks you dread: one park ranger turned a two-to-three-day paperwork process into a 45-minute tool that will collectively save 7,000 labor days across 430 parks.
- Cultivate diverse sources of inspiration—your unique experiences and inputs, combined with disciplined prompt variation, yield differential outputs from the same AI model.
Summary
Jeremy Utley opens by recalling Winston Churchill’s bathtub-dictated speech to illustrate the timeless nature of creativity’s unguarded moments and observes that modern professionals can now command an AI assistant with equal intimacy. As an adjunct professor at Stanford University’s design school, he has shifted from authoring a leading text on idea generation to immersing himself in generative artificial intelligence, learning alongside students and corporate teams about how this technology can augment human invention.
In his first chapter, Utley argues that users should ask an AI model to interrogate them first, requesting a sequence of questions that reveal workflows, objectives, and performance indicators. This “AI teaching itself” approach equips the system to offer two obvious and two non-obvious recommendations for integrating AI into daily responsibilities. He illustrates the point with a national park ranger who automated 2–3 days of paperwork into a 45-minute natural language tool, a solution that is projected to save 7,000 human labor days across 430 parks this year.
Chapter two contrasts two orientations toward AI: tool versus teammate. While AI can improve speed by 25 percent, output volume by 12 percent, and quality by 40 percent, less than 10 percent of professionals realize meaningful gains because they treat the model as a passive instrument. Outperformers, by contrast, coach their AI collaborators—providing feedback on mediocre results and prompting them to ask clarifying questions. Utley demonstrates how role-playing difficult conversations with AI can refine one’s approach, building psychological profiles and rehearsing dialogue.
In his final chapter, he revisits the definition of creativity as “doing more than the first thing you think of,” noting that generative AI makes it easier than ever to reach a “good-enough” answer. Yet achieving world-class innovation requires volume and variation in prompts, time to vet outputs, and the deliberate curation of inspirational inputs. According to Utley, creativity in the AI era demands disciplined exploration, leveraging personal experience and diverse influences to secure distinctive results. He concludes that rather than using AI, professionals must work with it as a collaborative partner to unlock unprecedented creative potential.
