Jeremy Utley, a Stanford professor and expert in innovation and design thinking, delivered a keynote focused on the intersection of idea generation ("idea flow") and generative AI, primarily ChatGPT.
He emphasized the importance of quantity and variation in creative ideation, illustrated with research and practical anecdotes.
The session included practical guidance for leveraging AI as a collaborator rather than an oracle, with live audience participation encouraged.
Key insights included productivity gains through AI, overcoming cognitive biases, and the necessity of a mindset shift for organizations and individuals adopting generative AI.
Action Items
(No specific due dates or names were provided for action items in the transcript. Please assign owners and deadlines as necessary.)
Principles of Innovation and Idea Generation
High volume and variation in ideas are essential to achieving quality outcomes; generating more ideas leads to better ideas.
Historical and cross-disciplinary research (e.g., Dr. Dean Keith Simonton, Jerry Uelman) demonstrates that aiming for quantity—even producing "bad" ideas—increases the likelihood of breakthrough innovations.
The "idea ratio" for commercial success: roughly 2,000 initial ideas are needed to arrive at one commercially viable solution.
Organizations and individuals must intentionally allow for bad ideas to unlock exceptional ones ("Dopey is the price of delight").
Adopting Generative AI for Creativity and Productivity
Generative AI like ChatGPT accelerates volume and variation in ideation, but its impact depends on user orientation and mindset.
The technology is rapidly evolving; previous negative experiences may be outdated due to significant improvements in AI capability.
AI should be treated as a dynamic collaborator, not as a search tool or oracle, to close the gap between AI’s potential and users' current productivity gains.
Research Insights: Human-AI Collaboration
Recent studies (including a Harvard Business Review publication) show that most teams using generative AI do not consistently outperform non-assisted teams.
Underperformance is often due to cognitive biases and the tendency to treat AI as an answer-delivering oracle rather than an iterative thought partner.
Teams that fully leverage AI treat it as a collaborative coworker, engaging in an iterative, conversational process.
Practical Tips for Effective AI Collaboration
Use natural language input, especially voice, for more productive and creative outputs—think and talk, don’t just type.
"Turn the tables": Prompt AI to ask clarifying questions before responding, enhancing understanding and relevance of outputs.
Regenerate responses multiple times; embrace the non-deterministic nature of generative AI for richer idea sets.
Provide feedback to AI, treating it like a junior collaborator. Iteratively guide and refine results rather than accepting the first answer.
Use AI to synthesize thoughts, draft memos, support follow-ups, and structure complex communications, especially when multitasking or ideating on the go.
Watch Outs and Common Pitfalls
First result bias: Unlike search engines, generative AI’s first output is not always the best—explore multiple responses.
Hallucinations (creative errors) are a normal, even useful, part of the creative process; randomness can spark innovation.
Generative AI is not like Siri or traditional assistants; it can process unsynthesized, rambling input and help users clarify and refine their thinking.
Driving Organizational Transformation with AI
Adopting AI is fundamentally a people transformation, not just a technology upgrade—HR, not IT, should drive upskilling and integration.
Sharing concrete use cases and examples within peer groups or organizations can unlock imagination and foster rapid adoption.
Audience Engagement and Live Practice
Attendees were invited to practice speaking to ChatGPT using voice input and to prompt the AI to ask clarifying questions before providing assistance.
Real-world examples highlighted how engaging with AI in low-stakes, personal contexts (e.g., parenting, cooking, family decisions) can spark broader application in professional settings.
Decisions
AI should be treated as a collaborator, not an oracle — Data and practical experience show higher creative and productivity returns when users engage AI as an iterative partner rather than a static answer provider.
Organizational AI adoption is a people transformation, not a technology implementation — Ownership should shift from IT to HR to foster skills, imagination, and use case sharing.
Open Questions / Follow-Ups
How can organizations best structure ongoing learning forums or peer groups to continually share AI use cases and spark imagination?
What support or training can HR provide to accelerate the adoption of conversational, collaborative AI practices across teams?