Only 16% of almost three dozen healthcare executives report having a policy for AI governance. At the same time, almost 20% say their institution has one or more committees of senior leaders from multiple departments set to oversee AI.

The latter finding “underscores the seriousness being placed on AI,” according to analysts at KLAS Research and the Center for Connected Medicine at UPMC in Pittsburgh.

The organizations interviewed 35 executives from 34 hospitals and health systems in November and December. This week they posted the results. Here are five of the report’s key questions, answers and observations.

1. Does your organization have a systemwide governance policy for AI usage and data access?

• 65%—No
• 19%—No, but we have broader policies that cover AI
• 16%—Yes

The report’s authors comment that, in addition to being in the early stages of AI adoption, several respondents also said their organizations are waiting for federal regulations to be issued before creating their own policies. More:

Other challenges or roadblocks to setting up system-wide AI policies cited by respondents included navigating the complexity of AI and its intersection with ethical, legal and compliance rules and policies; a lack of internal expertise to evaluate AI solutions in a clinical environment; a desire to focus on more-immediate priorities; and a need to first develop mature IT infrastructure to support AI applications.

2. Has your AI governance policy been updated in the last year?

• 55%—No
• 36%—Yes
• 9%—Policy has been updated but is not specific to AI

For those organizations that have established or are in the process of developing governance policies specific to AI, the policies “seek to oversee AI usage and address ethical, legal and security concerns,” KLAS and CCM note. More:

In many cases, the policies are developed by AI-specific committees tasked with ensuring AI is used responsibly, especially as it relates to patient data protection and regulatory compliance. Some organizations have formed AI committees but are still in the process of establishing AI policies.

3. Does your organization have policies in place that are specific to generative AI?

• 64%—No
• 11%—No, but policies are in process of being established
• 11%—Yes
• 7%—No, but there are verbal recommendations
• 4%—No, it's too early to have policies
• 4%—No, but a general governance process is used for generative AI

The authors here quote a vice president who remarks:

“The main policies are that AI is not to be used without human review and that AI is a tool to support people. But at the end of the day, regardless of whoever the clinical user is, they are responsible when they sign off on the AI documentation support or recommendation. If they are not comfortable with a recommendation, they need to be able to override it and explain why. If they are just blindly accepting things, then they should not be working for us.”

4. Does your organization plan to adopt AI that integrates with your EHR?

• 73%—Yes
• 17%—Unsure
• 10%—No

Most respondents agree that AI solutions will need to integrate within the EHR, the authors state. However:

They expect their organizations may end up using solutions from both EHR vendors and non-EHR vendors. Using both types of solutions would allow a health system to tailor AI to their unique needs and optimize ROI.

5. How does generative AI complement and/or enhance the work of your healthcare professionals?

• 26%—Improves efficiencies (general)
• 17%—Provides visibility into clinical decisions
• 11%— Automates repetitive tasks
• 11%—Educates & communicates with patients
• 9%—Improves documentation & charting
• 6%—Improves physician satisfaction
• 6%—Alleviates staffing shortages
• 3%—Reduces cognitive burden
• 3%—Improves radiology & pathology
• 3%—Improves patient throughput

Generative AI is “expected to automate routine tasks such as prior authorizations, scheduling and documentation,” the KLAS and CCM analysts point out. “This would then free up healthcare professionals to focus directly on patient care and allow organizations to optimize resource allocation.” They quote a hospital VP who observes:

“AI changes the user interface a lot, so it changes the way that people can interact with the record. It changes and accelerates the way we do data analysis.”

The full report is downloadable here.