Susan Ariel Aaronson is a Research Professor of International Affairs at the Elliott School of International Affairs, George Washington University (GWU). She is also a GWU Public Interest Technology Scholar and co-Principal Investigator of the NSF-NIST Trustworthy AI Institute for Law and Society. At GWU, she directs the Digital Trade and Data Governance Hub, and she is a Senior Fellow in Economics at the Center for International Governance Innovation (CIGI).

In her keynote, Professor Aaronson will address competition in AI, the rise of AI nationalism, and the risks of overcapacity in the sector. She will argue that governments, determined to invest substantial amounts of taxpayer money in AI, may unintentionally generate overcapacity, dumping of AI, and less effective regulation.

Questions from Grazia Errichiello

 

The theme of this year’s EAEPE conference refers to the “Janus face” of AI—its dual potential for empowerment and harm. How does this duality manifest in your own research and professional experience?

Aaronson: I see it every day in the US given the Trump Administration and Congress’s unwillingness to regulate either the data, the business models, the companies or the technology of AI. I also see it in my own research. I am speaking here about AI overcapacity – producing too much AI could lead to the dumping of AI without guardrails. I am also researching what companies say about responsible AI on their websites, how they train their chatbots (as revealed by technical documents) and do they care about responsible AI; their user agreements and then we ask chatbots about how they are trained. We see a lot of hypocrisy.

As scholars working at the intersection of public interest, international affairs, and the future of work, what do you believe are the most urgent societal questions we should be asking about AI today?

Aaronson: How can AI help us as scholars? How will AI change how we connect as humans and how we connect ideas? Can we resist AI? How can we use AI to increase human capacity and not substitute for it?

What are the major obstacles to developing global norms for ethical AI, and how can they be overcome in a multipolar, often fragmented world order?

Aaronson: No one knows what ethical AI looks like. Ethical AI is normative. I am not a fan of “ethical” AI. I want to see AI be more responsible and ethical but what that means in reality is unclear.

What risks do you see in the growing concentration of AI capabilities within a few dominant corporate and state actors, particularly for emerging economies and marginalized communities?

Aaronson: I don’t think this is true today, although it was last year. Many governments—for example, Switzerland most recently—have developed their own LLMs, which reduces the dominance of just a few actors.

How do you see your research contributing to the broader aims of EAEPE particularly in advancing institutional, heterodox approaches?

Aaronson: I hope my research contributes to our understanding of how institutions adapt to changing technologies and expectations of governance.

Continue reading the interview with Phoebe Moore
Or go back to the Editorial Page