Ai Webinar "The Creativity Ceiling": Where Human Brilliance Still Outshines 200,000 Ai Responses"
Abstract: This talk presents findings from a comprehensive large-scale empirical study that systematically compares the divergent creativity capabilities of humans and large language models (LLMs), including cutting-edge systems such as GPT-4 and DeepSeek-R1.
The research employs the Divergent Association Task (DAT), a rigorously validated and algorithmically scored creativity assessment tool, to analyze an extensive dataset comprising over 10,000 human participant responses and more than 200,000 model-generated outputs.
The results reveal that while human creativity scores are marginally higher than those of LLMs on average, humans demonstrate significantly greater variability in their creative performance. Most notably, humans exhibit substantially higher levels of originality at the upper tail of the creativity distribution—a critical area where current LLMs consistently underperform and fail to match human capabilities.
These comprehensive findings illuminate fundamental differences in the underlying generative processes and cognitive mechanisms between human creativity and artificial intelligence systems.
Bio: Professor Dawei Wang currently serves as an assistant professor at The University of Hong Kong Business School and holds a concurrent position as an external faculty member at the prestigious Northwestern Institute on Complex Systems (NICO) at Northwestern University.
He earned his Ph.D. in Management and Organizations from the renowned Kellogg School of Management at Northwestern University. Professor Wang's research agenda centers on human-AI collaboration, systematically exploring how artificial intelligence can function independently or cooperatively to enhance human performance across diverse domains including creativity, healthcare, and scientific research. His scholarly work uniquely integrates rigorous experimental methodologies with advanced computational approaches to examine the underlying mechanisms of effective human-AI teaming.
Through this interdisciplinary approach, he identifies optimal conditions under which AI can meaningfully complement human judgment, foster innovation, and enhance problem-solving capabilities, contributing valuable insights to the evolving landscape of human-machine collaboration and organizational behavior.