Generative artificial intelligence models not only make mistakes, but they also have the ability to spread misinformation, discriminate, or invade a person’s privacy. Policy initiatives seek to regulate these risks, often by framing them as risks to fundamental rights, democracy, or the rule of law. The EU AI Act, for example, requires that the risks associated with generative AI systems, including possible negative effects on human dignity, fundamental rights, and democratic processes, are identified and mitigated, and any residual risk is “acceptable”.
However, it is unclear how such aspirational statements can be translated into concrete, actionable rules. Any attempt to develop workable rules faces two connected challenges. First, it is unclear how the impact of generative AI on inherently ambivalent and context-dependent concepts such as dignity, privacy, or democracy can be measured. Second, even if it is possible to formulate appropriate benchmarks and measure the performance of generative AI models against these benchmarks, there is no established method to determine when the risks to protected interests posed by generative AI are acceptable, and when they are impermissibly high.
This talk explores both problems. It examines the role of generative AI in a democratic society and the value tensions that arise when policy makers require the mitigation of risks to a broad set of protected interests. It shows that some generative AI models necessarily involve a trade-off between indicators of performance that capture some of these protected interests, for example fairness and accuracy, and makes suggestions for how these trade-offs can be resolved.
No prior registration necessary to attend.
Prof. Carsten Gerner-Beuerle
Prof. Gerner-Beuerle joined the University College London (UCL) as a professor of commercial law in 2017. Prior to this, he worked at the London School of Economics, King’s College London and Humboldt University Berlin.
Carsten holds degrees in law and economics from Humboldt University Berlin (First and Second Legal State Exam, Ph.D.), the University of Minnesota (LL.M.) and the University of London (M.Sc. in Economics). He has also held visiting positions at various institutions in Europe and the United States.
He is a Global Distinguished Professor of Law at Notre Dame Law School and was previously visiting professor or visiting scholar at Duke University, the University of California at Berkeley, Trinity College Dublin, the Max Planck Institute for Research on Collective Goods, and Heidelberg University. Carsten is admitted to the bar in Germany and the United Kingdom, regularly advises a German law firm on matters of corporate law and corporate insolvency, and has prepared studies for the European Commission and the European Parliament on the reform of corporate governance, financial regulation, and private international law.
He is a research member of the European Corporate Governance Institute (ECGI). Currently, he is co-investigator on a Leverhulme-funded project concerning the regulation of AI.