Time: 10:30 AM EST
Date: June 25, 2024
From the EU AI Act to State regulations in the United States, companies in sectors such as insurance and human resources face new obligations to assess the fairness of AI solutions. And yet the data scientists and AI governance professionals responsible for meeting these requirements often lack the demographic data necessary to undertake accurate and comprehensive bias testing.In this webinar, a group of legal and technical experts will explore how practitioners can overcome this challenge.
The panel will discuss the implications of new regulatory proposals like the EU AI Act, which carves out an exception to the EU’s ban on collection of sensitive data for the purpose of bias mitigation; Colorado’s new insurance regulations, which would require insurers to adopt specific data imputation techniques; as well as the potential role of guidance from standards agencies and organizations like NIST, in resolving these tensions.
Experts will discuss:
- The impacts of demographic data gaps and relevant AI and privacy laws on bias assessments, with a focus on the insurance industry.
- State-of-the-art strategies for imputing demographic data based on correlations with other information.
- The pros and cons of data imputation vs data collection, including the legal, regulatory, and technical barriers to collecting demographic data.
- Guidance for policy makers and practitioners navigating this complex topic
Speakers:
- Miranda Bogen, Director, AI Governance Lab, Center for Democracy & Technology
- Reva Schwartz, Research Scientist/Principal Investigator for AI Bias, National Institute of Standards and Technology
- Myriah Jaworski, Attorney and Member, Data Privacy and Cybersecurity, Clark Hill PLC
- Chris Cooksey, FCAS, MAAA, CSPA, Senior Director of Advanced Analytics, Guidewire
- Philip Dawson, Head of AI Policy, Armilla AI