Erik Brynjolfsson is the director of the Stanford Digital Economy Lab and the Jerry Yang and Akiko Yamazaki Professor at Stanford’s Institute for Human-Centered Artificial Intelligence. His research examines how digital technologies affect productivity, work, business strategy and the economy. He is also a senior fellow at the Stanford Institute for Economic Policy Research.
Brynjolfsson holds bachelor’s and master’s degrees from Harvard University in applied mathematics and decision sciences and a doctorate from MIT in managerial economics. His work has examined the productivity contributions of information technology and the complementary importance of organisational capital, skills and other intangible investments.
With Andrew McAfee, he wrote The Second Machine Age and Machine, Platform, Crowd, exploring the implications of digital technology for work and value creation. These books connect changes in capability with the economic and organisational choices that influence their effects, rather than treating technological progress as an outcome independent of institutions or leadership.
His research with Danielle Li and Lindsey Raymond on generative AI at work studies assistance in a real customer-support setting. It finds differing effects across workers, making experience and context important to interpreting the results. That distinction matters for executive audiences: evidence from a defined application can inform a decision, but should not be converted into a guarantee for every workflow. His current research also examines transformative AI, economic measurement and the relationship between technology and prosperity.
For executive audiences, Brynjolfsson is relevant when an AI investment needs to be discussed through evidence and economics. A useful brief might concern why productivity takes time to appear, how work should change or which capabilities support a better result. His perspective encourages leaders to examine complementary investments and human contribution, keeping the business question broader than the performance of a model or the cost of a licence.
