Resource Center / Tools and Trackers / What AI Governance Leaders are Thinking About
What AI Governance Leaders are Thinking About
This resource presents prominent themes and questions identified by AI governance leaders from industry, government, academia and civil society.
Published: September 2023
In July 2023, the IAPP AI Governance Center convened a group of leaders from industry, government, academia and civil society in Portsmouth, New Hampshire to discuss the future of AI and responsible governance. What followed was a two-day discourse that covered everything from in-house AI governance structures to system bias testing to different approaches to AI governance regulation and enforcement. The panel conversations and use-case exercises produced not only robust discussion and debate, but also generated over 70 questions and comments from retreat participants. This resource presents some of the most prominent themes and questions identified through the discussions.
What AI Governance Leaders Are Thinking About
Auditing the Model
"When developing best practices to audit their AI models, how should organizations effectively govern algorithmic models, data sets and algorithmic parameters?"
Corporate Governance and Management
"If management of AI governance increasingly relies on other AI systems, how does that challenge traditional notions of policy being documentable, stable, accessible and auditable?"
Cultural Awareness
"What human cultural elements should organizations be aware of and address in developing AI?"
Data Governance
"How do you balance data set quality for issues like bias prevention and data science with data minimization requirements for effective AI governance?"
Intellectual Property
"How do regulators in different jurisdictions approach the impacts of AI on intellectual property concepts and laws?"
Law and Policy
"What are the operational considerations of a rights-based regulatory framework versus a risk-based framework for AI governance?"
"Should there be a single central AI regulating body, or should AI be governed by existing relevant, competent authorities?"
Liability
"Where does liability sit, noting the complex and multilayered technology stacks with developers, deployers, users or other parties for products that work with or use AI?"
Risks to the Individual
"How can organizations quantify risks to individuals when risks and technologies are constantly changing?"
Risks to Society
"Is the primary mechanism for addressing societal risks and potential harms by AI a solution that targets technical progress, regulations, consumer trust or something else?"
Values and Principles
"How do we choose one aspect of fairness or one ethical value without biasing the model in favor of one group’s preference?"