Designing AI Governance Structures

AI governance structures are the formal mechanisms through which oversight of AI is exercised in an organisation. They define decision-making processes, establish clear lines of responsibility and accountability, and create reporting and escalation pathways. By providing such guardrails, they give teams the confidence to safely innovate and can directly drive business value. 

Governance structures can take many forms and be given varying degrees of authority. For larger organisations, a common approach is often the creation of a centralised, dedicated AI governance body which consolidates cross-functional expertise in the organisation.

HTI’s AI Governance Snapshot, Designing AI Governance Structures, provides guidance on two key questions for organisations to consider when designing governance structures to manage AI systems.

Foundational questions for AI governance structures: 

1. Form of structure: Should your organisation establish a new, dedicated AI governance structure, or adapt existing structures to include AI oversight?
2. Location of authority: Should AI governance authority be centralised, decentralised, or hybrid?

As organisations increasingly look to AI for productivity and other benefits, those that successfully leverage AI governance structures will be better placed to manage AI’s risks and capture these opportunities.

“We’ve reached an inflection point when it comes to AI governance. Effective AI governance structures are no longer a nice-to-have. Standards and regulators have made it clear that organisations should assign accountability and have clear decision-making processes for AI systems. That is exactly how AI governance structures can help organisations.”

- Nicholas Davis, Co-Director of HTI