RESEARCH STATEMENT
My research interests are broadly in the application, innovation and impact of AI and frontier technologies on society and specifically in the future of work.
In particular, I am interested in:
- Societal consequences of generative AI
- Human creativity and creative markets in response to AI
I also have interests in:
- Multi-agent systems to model social processes
- Design and analysis of human-AI interfaces in healthcare
I apply a combination of deep learning, Large Language Models, multimodal feature extraction, and causal inference methods to large-scale datasets to derive insights from unstructured data, specializing in working with image data.
PUBLICATIONS
1. Generative Artificial Intelligence, Human Creativity, and Art
Eric B. Zhou, Dokyun Lee
Published at Proceedings of the National Academy of Sciences Nexus[PNAS Nexus] [SSRN] [CVPR 2024 AI Art Gallery announcement] [New Scientist] [Phys.org] [AAAS Press Release]
[Abstract]
"Recent artificial intelligence (AI) tools have demonstrated the ability to produce outputs traditionally considered creative. One such system is text-to-image generative AI (e.g., Midjourney, Stable Diffusion, DALL-E), which automates humans' artistic execution to generate digital artworks. Utilizing a dataset of over 4 million artworks from more than 50,000 unique users, our research shows that over time, text-to-image AI significantly enhances human creative productivity by 25% and increases the value as measured by the likelihood of receiving a favorite per view by 50%. While peak artwork content novelty, defined as focal subject matter and relations, increases over time, average content novelty declines, suggesting an expanding but inefficient idea space. Additionally, there is a consistent reduction in both peak and average visual novelty, captured by pixel-level stylistic elements. Importantly, AI-assisted artists who can successfully explore more novel ideas, regardless of their overall novelty prior to AI assistance, may produce artworks that their peers evaluate more favorably. Lastly, AI adoption decreased value capture (favorites earned) concentration among adopters. The results suggest that ideation and filtering are likely necessary skills in the text-to-image process, thus giving rise to "generative synesthesia" - the harmonious blending of human exploration and AI exploitation to discover new creative workflows."
[Conference Presentations]
Oct. 2023 |
INFORMS Annual Meeting 2023 |
Phoenix, Arizona |
Oct. 2023 |
INFORMS Workshop on Data Science 2023 |
Phoenix, Arizona |
Sep. 2023 |
Wharton Business & Generative AI Workshop |
San Francisco, California |
[Invited Talks]
Apr. 2024 |
Cornell University (virtual) |
PROJECTS
4. Creative Markets in the Age of Generative AI: Strategic Shifts and Labor Market Health
Eric B. Zhou, Dokyun Lee, Gordon Burtch, Daniel Rock, Prasanna Tambe
Analysis...
[Conference Presentations]
March 2025 |
Artificial Intelligence in Management (AIM) Conference |
Los Angeles, California |
3. Who Expands the Human Creative Frontier with Generative AI: Hiveminds or Maverick Masterminds?
Eric B. Zhou, Dokyun Lee, Bin Gu
Under review...
[Conference Presentations]
Dec. 2024 |
Conference on AI, ML, and Business Analytics |
New Haven, Connecticut |
Oct. 2024 |
Conference on Information Systems and Technology (CIST) |
Seattle, Washington |
Sep. 2024 |
Wharton Business & Generative AI Workshop |
San Francisco, California |
Aug. 2024 |
Academy of Management Annual Meeting 2024 |
Chicago, Illinois |
May 2024 |
Wharton AI and the Future of Work |
Philadelphia, Pennsylvania |
2. Reboot of: "Economics of Image-Based Seller Quality Signals"
Avery Chen,
Eric B. Zhou, Yingkang Xie
Analysis...
1. Economics of Image-Based Seller Quality Signals
Eric B. Zhou, Xiang Hui, Dokyun Lee
New version in progress...
[Abstract]
"In online marketplaces, sellers can rely on alternative mechanisms to signal their quality when they lack rich transaction histories. Using scraped data on GPU sales from eBay, we find that certain image signals can substitute for reputation to increase conversion rates amongst sellers with less than 100% positive reputation, and conditional on making a sale, can realize a 5% price premium on average. However, the effects are only significant for less reputable sellers."
[Conference Presentations]
Dec. 2022 |
WISE: Workshop on Information Systems and Economics |
Copenhagen, Denmark |
AWARDS
-
Marketing Science Institute research grant ($5,000)
-
Questrom Outstanding Research Award
-
Nominated for: Falling Walls Science Breakthrough of the Year 2024 in Art & Science
-
WISE 2022 Best Student Paper Finalist