An increasing number of large health systems are considering hiring chief AI officers. But what skill sets are most important in these roles and how can health systems set these executives up to succeed? Healthcare Innovation recently put these and other questions to two consultants at executive search firm WittKieffer. 

Sturman: There may be some indecision about whether this role is something everybody needs or not. Does everyone need a leader in AI? I think the simple answer is that it depends on an organization's sophistication and their level of innovation. Organizations that are a bit more strategic in the use of data and technology are leveraging AI in this new and innovative way, whereas there are still a lot of organizations saying we don't need to invest in an AI leader yet. 

We are seeing academic health systems look at AI from a perspective of not just research, but truly from a differentiation standpoint — how they can influence the behavior of their providers, make the burden less, while also affecting patient care. They have sophistication in their electronic health records, and this is translating to the next level of data usage, and that's really AI.

When they hire a chief AI officer, their goal is getting from a pilot to an enterprise AI program. That's inclusive of things like governance, change management, and advanced data definitions. 

HCI: When they don't have someone yet with this title, are they doing a lot of this work by committee — like a data governance committee. In that case, is it clear who's leading this work, or is it all over the place?

Durst: It’s all over the place. I can think of two academic health organizations where it is specifically done by committee. In other organizations, it's the data and analytics leader who may know enough about AI to lead it but isn't a specialist. In some organizations. it's the CMIO, who may have a bit of a data science background, who's taking that on.

Sturman: Even if someone's not recruiting for an AI leader, and they are recruiting for a CIO or CMIO, if AI is not part of their job description, then we would advise you that you're missing the boat. You can't implement an application today without the vendor talking about AI.

HCI: Zach, you said the number of people with this title has quadrupled in the last couple years. What does the reporting structure usually look like? Are they reporting to the CIO?

Durst: It is a small sample size so far, but the most common thing we're seeing is that they are reporting up through the CIO with a strong tie to the CMIO. In some cases the chief AI officer is a dyad partner to the chief data and analytics officer.

HCI: If a large academic medical center came to you and said we need a chief AI officer, can you generalize about what skill set or background they're looking for to fill that role?

Sturman: There are a couple of buckets they tend to fall into. One is an industry leader that's looking to make a change into a mission-oriented environment — so think master’s or Ph.D.-trained AI leaders who are currently doing this for financial services or commercial or tech, and they're drawn to the mission of healthcare. They want to come in and make a difference, and they're going to bring that high-level technical acumen to the work.

Durst: Those that come from healthcare might have a background in informatics or be a physician with data science experience.

Sturman: A physician leader is really coming in from the perspective of clinical change management. They are thinking if we are changing the workflow to include AI, how do we make sure our clinical community is going to trust what the AI is telling them? 

Durst: Rarer still are data and analytics people in healthcare today who have really advanced their skill set to a point where they have either solely focused on AI or they are looking at AI in addition to a broader data analytics portfolio.

HCI: In imaging, AI has been around a little longer, but perhaps operating a little bit separately. So if you’re hiring a chief AI officer, are they going to oversee all that work, too?

Sturman: We’ve seen a convergence now toward what I'll call platform-based AI, not necessarily one-off solutions. So even the imaging solution vendors have generalized their platforms to be not just imaging-based solutions to be much more broad in clinical focus. I think we've seen that the AI leader needs to understand clinical workflows, imaging workflows, revenue cycle workflows — the whole gamut of how really AI is going to influence what I'll call operations strategy and patient care.