In episode 13 of Power and Consequences, we (Gary Gensler and Simon Johnson) explore whether Artificial Intelligence can be expected to behave “responsibly”.
When we started researching and teaching about AI in the late 2010s, there were policy debates about AI ethics or what later came to be called “responsible AI.” In those years, as machine learning was increasingly used in decisions about who gets opportunities in our economy — e.g., consumer loans, entry to college, employment, healthcare, or insurance — responsible AI was mostly about the accuracy, bias, and explainability of the algorithms. Given AI’s insatiable use of data there also were concerns about data privacy and surveillance.
As AI capabilities have increased since the release of ChatGPT in late 2022, public expectations — and unease — have expanded, leading to broader definitions of what “responsible” AI might mean moving forward. There is now growing concern about deepfakes and intimate imagery, as well as about AI’s impact on mental health, on children, and education more broadly. The impact on jobs also looms large.
The public also is increasingly concerned about the business models of AI companies, including the boom in data center investment, effects on electricity prices, and the potential for AI agents to run amok. While numerous U.S. states have enacted AI related laws, many in the AI industry have resisted the imposition of meaningful federal legal safeguards, but public opinion may now be shifting. Further, companies using the preview of Anthropic Claude Mythos coding model have uncovered thousands of bugs in their own software — highlighting the real risks of threat actors using AI to disrupt critical parts of our economy.
What is the right public policy response? How is it best to both promote the public good and innovation? Are there ways to steer AI development in a better direction? What exactly would that involve? And how does the technology race with China play into these considerations?
Gary has written and spoken extensively about a wide range of analytical, business, and public policy issues in the context of AI. Here are some of his most relevant items (listed in reverse chronological order by medium):
Podcasts
‘On AI in Finance’, EconoFact podcast with Micheal Klein, May 3, 2026
‘Episode Seven: Artificial Intelligence’, Power and Consequences podcast with Simon, April 14, 2026 (i.e., part of our podcast series, within which we expect to have many discussions of AI).
‘Trump, Trade, and AI Growth’, VoxTalks Economics podcast, January 2026
Writings
‘Artificial Intelligence Development and Policy Landscape’, December 2025, published as part of our co-edited CEPR volume, The Economic Consequences of the Second Trump Administration.
‘AI, Finance, Movies, and the Law’, Yale Law School Remarks, February 13, 2024
‘Isaac Newton to AI’, National Press Club Remarks, July 17, 2023
‘Deep Learning and Financial Stability’, with Lily Bailey, November 2020, which might seem like a long time ago but still has quite a bit of relevance.
Short Videos (3 -4 minutes each), when Gary was Chair of the Securities and Exchange Commission: part of Office Hours with Gary Gensler. Released February-October 2024.
AI, Investors, Issuers, & the Markets
Systemic Risk in Artificial Intelligence
Fraud and Deception in Artificial Intelligence
Conflicts of Interest in Artificial Intelligence
Simon also has written about AI, for example in Power and Progress: Our 1,000 Year Struggle Over Technology and Prosperity, with Daron Acemoglu, which attempts to put this latest technological transformation in broad historical context (actually more than 1000 years!). The mini-comic version of this book, which emphasizes the issues through the lens of what AI may bring, is available for free via the MIT Stone Center for Inequality and Shaping the Future of Work.
And here is a link to Geoffrey Hinton’s Nobel banquet speech from December 2024 discussed in this podcast. It is chilling. Everyone should read it.
Note: In this episode we don’t focus on the key issue of how AI will impact jobs. We covered this in part of episode seven of Power and Consequences (linked above), and plan to pick this topic up in a further episode (or more than one!) soon.
