How to design an AI ethics board

Clear responsibilities are foundational: The paper identifies that AI ethics boards need well-defined responsibilities across five key design dimensions—from their scope of authority to their legal structure—because vague mandates have contributed to previous board failures in the industry.

Board composition directly impacts effectiveness: Who sits on the board matters significantly; the paper suggests that board member selection affects the organization's ability to identify and mitigate AI risks, implying that diverse expertise and perspectives are necessary for meaningful oversight.

Decision-making authority must be explicit: The paper examines whether board decisions should be binding versus advisory, indicating that ambiguity about enforcement power has been a practical challenge that undermines boards' ability to actually reduce AI risks in organizations.

Adequate resourcing is often overlooked: The paper identifies resource allocation as a critical design choice, suggesting that under-resourced ethics boards cannot effectively fulfill their risk-reduction mandate—a practical constraint that organizations frequently underestimate.

Design choices form an interconnected system: Rather than offering a single "best" approach, the paper presents multiple options across five dimensions and discusses how different combinations affect risk reduction, recognizing that effective board design depends on how these choices work together within specific organizational contexts.