AI Boosts Creativity Only at Moderate Use, Studies Find
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Hsuan-Che Brad Huang, a PhD researcher at the University of British Columbia, published findings showing AI use enhances creativity only at moderate levels of engagement—the 'Goldilocks zone'—with too little or too much use diminishing original output
- Huang's first experiment tested roughly 150 participants on building an impromptu business for a student named 'Mike' with $10 in seed funding, comparing groups prompted to use ChatGPT once, four-to-six times, or nine-plus times; the four-to-six prompt group scored highest on novelty, utility and business value
- A second experiment with 319 participants produced the same inverted-U pattern, and a survey of fashion designers, visual artists, authors, animators, technologists and influencers found that bosses rated moderate AI users (scoring 4-5 on a 7-point scale) as the most creative
- Columnist David Robson tested the thesis himself, using ChatGPT to develop a 30-minute movie outline about an engagement-party memory card that exposes a family corruption scandal, and concluded the AI broadened his options but required his own judgment to maintain narrative coherence
- Grace Liu at Carnegie Mellon University has a peer-reviewed-pending study showing people quickly become reliant on AI and less persistent in its absence—a concern the columnist echoes for long-term cognitive development
- The article's broader framing distinguishes AI as a useful perspective-shifter from a replacement for the 'sensation of synapses straining,' arguing that outsourcing thinking forfeits the surprise of ideas emerging from personal experience
Why it matters: The finding reframes the AI-and-creativity debate from 'whether to use it' to a calibration question, with three converging studies—150-person lab, 319-person replication, and a survey of working creatives whose bosses rated them—all pointing to the same moderate-use optimum. Heavy AI users risk weakened problem-solving persistence, per Carnegie Mellon's Grace Liu, a tradeoff that matters most for early-career creatives building their craft.
