Meet the newest member of your team: the chief AI officer

Oklahoma turned heads late last year with a report that called for creating a new C-suite position, a chief artificial intelligence officer.

The report, authored by a governor-formed task force on emerging technologies, said the role would be “pivotal” in overseeing implementation of the technology across state government. It further defined the CAIO role by laying out a candidate’s requisite experience, including a STEM degree, at least 10 years of experience in developing AI systems and innovation programs, and a background in developing budgets for emerging technology and AI risk management programs.

The task force also made clear the CAIO’s duties. The officeholder would be the “senior advisor for AI” to the state’s executive branch; chair a new AI oversight committee; lead the way on AI strategy and policy; coordinate with AI experts within each agency; maintain an inventory of AI tools; scale AI uses; create communications plans to disseminate information; and oversee any state AI task forces.

Oklahoma’s suggested CAIO hire comes on the heels of President Joe Biden’s AI executive order last year, which mandated that every federal agency hire or designate a CAIO by year’s end. The Department of Justice, for example, recently announced that Jonathan Mayer would serve as its first chief science and technology advisor and chief AI officer.

Mayer’s elevation, plus the likely hiring of other federal CAIOs and the Oklahoma report’s recommendations, might make state and local governments think they need to hire their own CAIOs to keep pace with their peers. But given the sheer number of C-suite officeholders in government, many experts caution against rushing out to fill the role.

“We're going through a phase of many chiefs,” said Alan Shark, executive director at the Public Technology Institute, which offers IT training to local governments. “Every time we have a new topic, instead of adding responsibilities [to an existing chief] and having a clear hierarchy, we [create new positions]. We have a chief data officer, chief ethics officer, chief technology officer, chief innovation officer, chief geographical officer.”

A CAIO for a state or local government could need a lot of skills, Shark further cautioned. The duties of a CAIO, like those outlined in the Oklahoma report, would require someone with a unique and broad skill set that could be hard to find in just one person.

Asha Palmer, senior vice president of compliance solutions at Skillsoft, agreed, adding that it will be almost impossible to have blanket guidance on what is required of a CAIO.

“As we think about what this CAIO role looks like, the reality is that it is going to have to be industry specific, it's going to have to be company specific,” Palmer said. “Depending on where the first regulations or guidance is issued, it may even have to be geographically specific.”

What’s more, some skills sought may have more urgency than others, she said. Risk management could be crucial given how concerned people are about the negative effects of AI.

“The concern around AI is about what could go wrong,” she said. “Risk managers are professionals at strategizing as to what could go wrong, and then what controls do you put in place so that you mitigate the risk of that going wrong.”

But no matter what the position ends up looking like, given the role AI is expected to play, there needs to be someone leading the way, whether they are called a CAIO or something else entirely, Shark said. He noted that Washington has placed much of the onus on its CIO to drive AI policy, while New Jersey Gov. Phil Murphy appointed former Chief Innovation Officer Beth Noveck as the state’s first Chief AI Strategist.

“AI is so pervasive, clearly somebody in every government unit has to be in charge of coordinating, thinking, training, trying,” he said.

Rather than have one person with the responsibility of overseeing everything related to AI in government, Palmer and others have suggested that a cross-functional team might be a better way to share the workload. That could take the form of a task force or committee and include people with expertise in technology, policy, ethics, risk management and workforce development.

A recent study from McKinsey & Company said governments could adopt such a multidisciplinary, cross-functional team to help establish guardrails and issue guidance on AI adoption and risk. That team needs to start early when planning for future AI uses, the report said, and needs more than just technologists. For a specific use case, the team also needs people from the agency who will be implementing AI, as they are best placed to know how it can work.

“You need to think through the use case, and who are all the stakeholders that need to be involved, to make sure that you're fixing the right problem using this technology, and not blindly applying the technology,” said Sameer Chowdhary, a partner at McKinsey and coauthor of the report.

Palmer agreed that identifying use cases is the best place for governments to start before hiring a CAIO or standing up a task force to implement AI. She noted, however, that the technology is already in use in various forms, so “catastrophizing” about the future may be unnecessary. Having someone, or people, in charge of AI implementation could help dispel any fears, she said. That is especially necessary in communicating with the public.

“Look back and understand how we've already been using it, where it's been deployed, its use cases that are currently being used and quite frankly, haven't been governed,” she said. “How has government used it in the past, and think about, what are ways to accelerate the good in that? What are ways to guard against the potential bad in that use case? All those things, I think, will help us guide our way forward.”