“Economics & AI” Fireside Chat: Professor Susan Athey and Dean Jon Levin

Susan Athey's Career

• Susan Athey is an economist and professor at Stanford Graduate School of Business (GSB).
• She became interested in economics while working as a work-study student for an economics professor and seeing how economic analysis could be used to change policies.
• As a PhD student at the GSB, she focused on auctions and market design, combining economic models with data to answer real-world problems.
• After teaching at MIT, she returned to Harvard and won the John Bates Clark Medal.
• She then made an unusual move by joining Microsoft as their Chief Economist, where she faced the challenge of predicting industry structures for new or non-existent industries and understanding the value of data.
• Despite the initial challenges, she found the work intellectually stimulating and had a significant influence on the company's decisions.

Susan Athey's Research and Teaching

• Athey emphasizes the value of rigorous economic analysis and being able to predict future market structures.
• She describes her role in convincing Microsoft to build Azure, despite initial skepticism, based on economic analysis and understanding of the industry structure.
• She mentions her work in promoting machine learning in economics and social sciences, which was initially overlooked but has now become a significant part of research.
• The speaker developed tools and software to connect machine learning techniques with traditional business analytics and economics research, enabling personalized policy design and analysis of causal relationships.
• She helped establish the Human-Centered AI Institute at Stanford to promote interdisciplinary research and collaboration between engineers, ethicists, and social scientists in developing and deploying AI systems responsibly.

Susan Athey's Work at the Department of Justice

• The speaker's work at the Department of Justice involved modernizing merger guidelines by incorporating data analysis and addressing scenarios involving companies with monopoly power.
• The revised merger guidelines consider how mergers affect not only today's customers but also future customers who may lose out due to potential monopolies.
• The DOJ is building a data science and technology team to better understand and analyze the tech industry, as it becomes increasingly important and complex, especially with the rise of AI.

Challenges and Opportunities in AI Competition

• There are concerns that dominant firms in AI may buy out competitors, create exclusives, and further entrench their market positions, leading to worse problems than those currently faced.
• However, there are also positive developments, such as the use of open-source large language models and retrieval-augmented generation, which make AI more accessible and potentially competitive.
• The future of AI competition is uncertain, with risks such as bottlenecks and the potential disappearance of open-source models, but also opportunities for increased competitiveness.

Building and Maintaining a Team of Engineers and Data Scientists in the Government

• The speaker discusses the challenges of building and maintaining a team of engineers and data scientists in the government.
• Despite the difficulties, the speaker highlights the passion and dedication of government employees, particularly younger individuals who are drawn to the mission-driven nature of public service.

The Importance of Technology Policy

• The speaker emphasizes the importance of technology policy and the need for individuals with expertise in this field.

AI and Competition

• The speaker expresses concerns about the potential impact of AI on competition, including the risk of collusion and information sharing that could lead to higher prices.
• The speaker suggests that the US government should consider developing its own open-source language model utility to compete with companies like OpenAI and Meta.

Managing Suppliers and Buyers in Marketplaces

• The speaker draws on her experience with marketplace companies to discuss the challenges of managing suppliers and buyers and the importance of understanding economics when building and growing a marketplace.
• Marketplaces often make the mistake of neglecting the supply side of their platform because it's hard to measure the ROI of nurturing suppliers.
• Marketplaces should focus on building tools and features that make it easier for suppliers to manage their businesses, even if the ROI is not immediately apparent.
• AI is emerging as a powerful tool that can help marketplaces improve the supplier experience and make it easier for suppliers to manage their businesses.