
February 19, 2026 by Melissa De Witte, Stanford University
Collected at: https://techxplore.com/news/2026-02-thinker-ai-creativity.html
Stanford d.school’s Jeremy Utley wants people to stop using AI. Instead, he wants them to work with it. “If you’re ‘using’ AI, I know you’re misusing it,” said Utley, an adjunct professor at the Hasso Plattner Institute of Design (aka the “d.school”). Utley argues that people fall into two categories when it comes to AI: underperformers who treat it like a tool and outperformers who treat it like a teammate.
“For a lot of people, that becomes a lightbulb moment that shifts their whole trajectory,” Utley explained.
Last year, Utley’s video interview, “How Stanford Teaches AI-Powered Creativity in Just 13 Minutes,” which has 2 million views and counting, made a compelling case: People are using AI wrong, and thinking about it the wrong way, too. “I’ve never gone viral as a professional nerd, so that was kind of a wild moment,” Utley said.
When ChatGPT launched in 2022, Utley, like many others, was stunned to see how the chatbot could produce in seconds (not without flaws, of course) what would have taken him hours. But then his design thinking instincts kicked in. Here was a creative partner for iteration in the design thinking process, one that could help him better understand his own creative vision.
Take composing an email. An underperformer asks AI to write a reply and accept what is most likely a very adequate response. But when you treat AI as a colleague, you invite collaboration. “You give it permission to ask you questions,” Utley explains. A prompt would be: “Help me respond to this email. Ask me any questions you need to get context.”
For Utley, that’s where creativity begins.
The technique, called reverse prompting, helps overcome the cognitive traps that limit innovation. “When humans are solving problems, they get lazy,” Utley explained. “They hit good enough and stop.” This is known as “satisficing,” a term coined by Nobel laureate Herbert Simon in 1956 to describe how people tend to settle for good enough, rather than push for better.

Jeremy Utley. Credit: Stanford University
ChatGPT and other generative AI chatbots can actually enable this tendency, Utley says, making it easier than ever to settle for an adequate answer.
The over-the-top flattery characteristic of some chatbots, what Utley calls “glazing,” can also be leveraged to promote creativity. One of the main principles of design thinking is ideation, a process that aims to generate as many ideas as possible without judgment. That sycophantic praise can create the psychological conditions for breakthrough thinking, said Utley, citing neuroscience research from Dr. Charles Limb at Johns Hopkins, who found that during creative flow states, the judgment center of our brain deactivates, with profound implications for creativity.
But when large language models like ChatGPT tell you all of your ideas are wonderful, you may feel less inhibited or constrained about proposing more ideas. As Utley put it in a recent blog post: “To enter into creative flow, you’ve got to turn off censorship, that critical part of your brain that says, ‘that’s a dumb idea.'”
In a framework Utley developed called “Stop Fighting AI Glazing,” he suggests two separate workflows: one focused on divergent thinking that provides psychological safety to explore wild ideas, and another that rigorously evaluates against predetermined criteria. “The key is knowing which phase you’re in: when you need volume and when you need rigor,” he said.
For example, when Utley is in the ideation phase of drafting headlines for a blog post, he’ll ask AI to come up with 20 options. “Most people don’t think to generate volume,” Utley said. But he doesn’t stop there. If he likes options 2 and 14, he will ask for more like them, providing the model with design constraints. But crucially, Utley will also explain why he likes those options. Maybe he likes the pun in option 2 or the alliteration in number 14.
“If you’ve never articulated what makes your work yours, AI will give you average. But if you’ve done the work to know yourself as a creative? AI becomes an extension of your voice, not a replacement for it.”
The real magic, Utley said, comes from teaching AI who you are, including: “your taste, your voice, your criteria for what ‘good’ looks like.” And the more specific you can be, the closer AI’s response will sound to your own tone and style. “AI amplifies your underlying humanity,” he said.
Staying ‘wildly engaged’ in the creative process
But what about other forms of discovery encountered in real life: the serendipitous passage in a book that transforms a research project, or the “aha” moment born from the difficult process of creation, which is itself a mode of learning? What happens when that’s outsourced to AI? As one widely publicized study from MIT showed, relying too much on AI may affect memory and retention.
Utley emphasizes that working creatively with AI does not mean conceding agency to the machine. When done skillfully, he says, it’s more like having a thought partner in the process, one that helps him move past what he calls “cognitive bottlenecks” and pushes him to think harder and more creatively. “Is the human catatonic or is the human wildly engaged in their life and practice?” Utley asks, noting he still reads books and does his own research.
“AI is about expanding the surface area of possibility,” he said.
Utley’s 5-minute emotional decision drill
Utley sees how AI can spur solutions to many kinds of questions, including emotional ones. For example, think of an emotional decision you’re wrestling with right now—anything you’d talk to a best friend about. Career crossroads, a difficult conversation you’re avoiding, or whether to take on a new project.
Open ChatGPT and say, “I’m wrestling with a decision. Before you give me any advice, please ask me three questions to understand my situation better.”
Then just talk to it. Answer the questions. Let it ask follow-ups.
That 5-minute exercise shows you what AI collaboration actually feels like—and it’s nothing like a Google search. “Most people walk away going, ‘Oh, that’s what he meant by teammate,'” Utley said.

Leave a Reply