Artificial intelligence has the potential to become a creative partner for researchers. Bots might be useful for triggering and batting around ideas, visualizing concepts in fresh ways or exploring information in different domains. However, in practice, the opposite is often true — bots flatten creativity.
Can AI be truly creative?
As a researcher who has spent 30 years studying human creativity, I think I know why. When a scientist feeds a data set into a bot and says “give me hypotheses to test”, they are asking the bot to be the creator, not a creative partner. Humans tend to defer to ideas produced by bots, assuming that the bot’s knowledge exceeds their own. And, when they do, they end up exploring fewer avenues for possible solutions to their problem.
The answer, in my view, is not to swerve away from using AI. Instead, researchers should learn how to use bots to boost their creativity.
I’ve been working with a test of rapid creative thinking called the divergent association task (DAT), which I think points to a way forwards. Participants have four minutes to come up with ten nouns that differ from each other semantically as much as possible. This is a hard task. Once someone thinks of ‘queen’, for example, they tend to call to mind related words such as king or knight — a phenomenon known as thought anchoring.
Most participants assume that if they ask a bot to perform this test, it will outperform them. But my colleagues and I have found that, on average, bot and human scores are about the same (D. Wang et al. Nature Hum. Behav. https://doi.org/hbhxbp; 2025).
Does using ChatGPT change your brain activity? Study sparks debate
By contrast, preliminary follow-up work shows that if participants ask the bot for a process with which to improve their performance, their own scores skyrocket. The bot proposes two steps. First, think of ten word categories — transportation, electronics, mammals and so on. Second, pick one word from each category. This process diminishes thought anchoring and preserves diversity of thought across participants, who each choose different word categories. The innovator and the bot are now partnering.
This experiment suggests that, when collaborating with bots, people should ask not what to think, but how to think. Work on human creativity backs up this idea.
Human creativity is often a concoction of blind alleys, detours and serendipity. But beneath the apparent chaos, studies have revealed some general principles for maximizing creativity in groups.
