Measuring What Matters: Understanding the Human Impact of AI

AI systems are tested and evaluated more extensively than ever, but the effects of AI on humanity are still not fully understood, while narratives about our future vacillate between apocalypse and utopia. Researchers are beginning to measure AI’s effects on people, and labs are increasingly expected to report on the impact of their own products.

What is missing is a common way to define, measure, and verify human impact so the findings can be trusted, whether produced in a lab, by a company, or by a public agency. This cross-sector conversation brings together Dr. Amit Goldenberg, Associate Professor at Harvard Business School, whose research examines how AI is shaping emotion and social connection; Dr. Avijit Ghosh, Lead Technical AI Policy Researcher at Hugging Face, who works on what current evaluation methods can and cannot capture; and Jeff Dunn, Chief Operating Officer of the Digital Trust Council, who handles the operational reality of being accountable for a product’s impact on millions of lives.

Moderated by Alex Pascal, Executive Director of the Berkman Klein Center for Internet & Society, the panel will take up what it would mean to define human impact in terms that hold significance across contexts, what kind of collaboration across research, industry, and government that would require, and how such an approach might inform a standard that holds AI to human values.

This panel is a public event within the AI Trust Summit, co-sponsored with the Digital Trust Council.