What should you call your organisation’s AI oversight function? An exercise in great power coming with great responsibility

Oversight is crucial to an organisation’s well-being by acting as a safety mechanism. Increasingly, organisations will need some form of oversight regarding their use of AI. Effective oversight can be envisioned in several ways, aiming for ethical AI, responsible AI, governed AI, compliant AI, and more. The adjective we use before AI shapes the structures and people we will likely hire to achieve the desired outcome.

An organisation aiming for ethical AI might recruit individuals with backgrounds in ethics, including academic researchers, ethicists, and advocates for justice and inclusion within AI. This function could be integrated into existing ethics boards, potentially reporting to the Head of Research & Development or a similar role.

For governed AI, organisations might align AI oversight with broader governance programs, incorporating it into the responsibilities of the Chief Information Officer, Chief Operating Officer, or even the Board, depending on the governance setup. They would likely seek individuals who understand corporate governance, programme creation, and oversight roles.

An organisation pursuing compliant AI will likely embed this within their compliance functions, possibly under the Chief Operating Officer, Chief Counsel, or Chief Risk Officer, depending on the industry. These organisations will probably prefer individuals with traditional compliance backgrounds who can interpret legislation and regulations and translate these into requirements for technical teams.

At Microsoft, we use the term responsible AI, with our Office of Responsible AI ultimately reporting to the Board. As a company that consumes, creates, and enables others to use AI, our situation is more complex than many organisations will face. We have more accountability. Responsible AI in our context is multifaceted and inherently multidisciplinary, involving professionals from various backgrounds working together to achieve ethical, governed, and compliant outcomes. Our recent Microsoft Responsible AI Transparency Report provides insight into our governance efforts and structure, particularly on page 27.

The terminology we use signals our goals and indicates the likely reporting line within your organisation. This, in turn, influences broader incentives, priorities, hiring preferences, and the overall shaping of this organisational capability. The individual leading this initiative holds great power and, therefore, great responsibility to the organisation, its employees, and its customers. Choose wisely.