According to Ethan Mollick on X, a new peer-reviewed study reports that large language models consistently generate more creative product development ideas than human participants recruited on Prolific, and that newer, larger models outperform prior generations; the paper also tests a creativity-boosting intervention that improves human ideation but does not enhance LLM creativity (as reported by Ethan Mollick citing the study). According to the study authors, model size and recency correlate with higher novelty and usefulness scores in expert ratings, indicating measurable gains in creative performance for product ideation compared to human baselines (according to the paper shared by Ethan Mollick). For businesses, this implies immediate opportunities to integrate state-of-the-art LLMs into front-end innovation workflows—idea generation, concept variation, and rapid product discovery—while human-targeted creativity training may not translate into LLM gains, suggesting dedicated prompt strategies and model selection are more impactful (as reported by Ethan Mollick summarizing the study’s findings).