Job Market Paper
Does Generative AI Crowd Out Social Learning? Evidence from a Field Experiment with Entrepreneurs
Recent experiments demonstrate that advice can significantly improve firm performance, yet many entrepreneurs lack access to knowledgeable advisors. Generative AI could democratize access to high-quality advice, but it also poses risks: if AI replaces learning from peers, mentors, and other advisors, the net effect on entrepreneurial performance could be negative, even if AI offers valuable guidance. I explored this through a nine-month randomized field experiment with 1,200 Kenyan small- and medium-sized business entrepreneurs. Participants were randomly assigned to a control group or to receive (1) a generative AI business assistant, (2) a mentorship program that offered advice and fostered peer learning, or (3) both AI and the mentorship program. Consistent with the idea that AI assistance can crowd out human advice, the AI business assistant caused entrepreneurs to discuss their businesses with fewer people and produced a 15% decline in profits and revenue compared to the control group. In contrast, entrepreneurs in the mentorship program received more advice from peers and experts and saw an 11% increase in profits and revenue. For those receiving both interventions, the mentorship program offset the network declines observed in the AI-only group, enabling entrepreneurs to capitalize on AI advice and resulting in a 27% increase in business performance. These findings suggest that while AI can harm performance if it erodes the social structures that support entrepreneurial success, with the right human support it can significantly improve entrepreneurial performance.
Working Papers
The Uneven Impact of Generative AI on Entrepreneurial Performance [ PDF ]
Nicholas Otis, Rowan Clarke, Solène Delecourt, David Holtz, and Rembrand Koning
Revise and Resubmit at Management Science
- First Prize, Wharton People Analytics White Paper Competition (2024)
- AOM STR Division Best Paper Award in Industry, Competition, and Strategic Entrepreneurship (2024)
- Finalist for the SMS Annual Conference Best Paper Prize Competition (2024)
- Nominated for the AOM Carolyn Dexter Award by the STR Division (2024)
-
Press:
VoxDev (Podcast);
Berkeley Haas News;
Charter
Gender Gaps and Generative AI
Nicholas Otis Katelyn Cranney, Solène Delecourt, and Rembrand Koning
Under review at Science
Peer-Reviewed Publications
Toward a Science of Delivering Aid with Dignity: Experimental Evidence and Local Forecasts from Kenya
Catherine Thomas, Nicholas Otis, Justin Abraham, Hazel Markus, and Gregory Walton
Proceedings of the National Academy of Sciences, 117, no. 27 (2020): 15546-15553
[
PDF
][
Appendix
][
Data+code
]