The corporate obsession with appointing an executive Chief AI Officer

You’re a Chief AI Officer! And you! And you too!” – Recently, requiring AI officers has become all the rage. (Funnily enough, there were Chief Internet Officers at one point, and there may still be vestiges of the role of Chief Blockchain Officer in some places)

In the US, the White House Office of Management and Budget (OMB) has now issued its first government-wide policy to “mitigate risks of artificial intelligence (AI) and harness its benefits”. In accordance with a previous Executive Order, this policy notably states that each US federal government “agency must designate a Chief AI Officer (CAIO) within 60 days of [today, 28 March 2024]”.

Fortunately perhaps for their budgets and hiring processes, “[a]gencies may
choose to designate an existing official, such as a Chief Information Officer (CIO), Chief Data Officer (CDO), Chief Technology Officer, or similar official with relevant or complementary authorities and responsibilities, provided they have significant expertise in AI and meet the other requirements”.

Now, I am all for thinking carefully about AIgovernance, but in most organisations I have found that AI governance is best handled as a team effort. Unfortunately, these CAIOs are responsible for the following, under this OMB policy:
– Coordinating Agency Use of AI
– Promoting AI Innovation
– Managing Risks from the Use of AI
The RACI matrix for these positions will not look very pleasant, in particular when combined with their current roles.

[Interestingly, no such clamouring for CAIOs the past decade when “mere” hyper-advanced machine learning and deep learning tools were deployed for many other use cases than GenerativeAI]

Some thoughts then:

– “AI” represents a broad range of technologies. Some variants present some risks, others present others. Don’t assume the risk is the same for all.

– AI governance involves many stakeholders, some of whom should be Accountable or Responsible for different parts of the overall strategy and action points.

– An “AI Officer” might not be the best strategy for every organisation, in particular because this creates issues of hierarchy. For instance, who is first consulted on the data protection or privacy aspects of AI, a DPO or a CAIO? If the CAIO is merely a coordinator between existing functions, why the title?

Much better to avoid labelling everything in advance and to start by building a framework that works for the organisation – and then seeing what additional roles this might require.

That has worked much better on the AI governance projects I have been asked to help out with.

Link to the OMB policy announcement: https://lnkd.in/emDPYqtE