Report: Health care AI adoption ramps up, but few pilots reach implementation

Health care leaders are excited about the promise of artificial intelligence (AI) to reshape the sector, but relatively few pilots of the emerging technology have been fully implemented so far, according to a report by Bessemer Venture Partners, Amazon Web Services and Bain & Company. 

Nearly all survey respondents said generative AI, which can create new content like images and text, will be transformative, while 85% of leaders at provider organizations said it could change clinical decision-making in the next three to five years — even with health care’s complex regulatory environment and low tolerance for mistakes. 

Still, only half of organizations have a clear AI strategy, while nearly 60% have created an AI governance committee. Meanwhile just 30% of AI pilots have reached the production stage, as executives struggle with security and data challenges, as well as integration costs and limited in-house expertise, according to the survey. 

Buzz surrounding AI in health care has increased in recent years, boosted by hopes the technology could help solve some of the industry’s most pernicious problems, like provider shortages, clinician burnout and a heavy burden of administrative work. 

There are plenty of jobs AI could take on, according to the report, which surveyed more than 400 leaders at payer, provider and pharmaceutical companies.  

The technology could help schedule patient appointments or worker shifts, support prior authorization requests, help document care and assist members as they enroll in health plans, among other tasks, executives said.  

Eighty-four percent of survey respondents across all segments of the health care sector believe AI will have an impact on clinical decisions, while 80% said the technology could reduce labor costs. 

Still, many AI projects are in the early stages of development, with nearly half in the ideation or proof of concept phases, according to the report. Provider organizations are more commonly moving their pilots into production: About 35% of their proof of concepts are implemented, compared with 31% among payers and 24% for pharma companies.  

“This wave of AI adoption has been driven by ‘test and learn’ urgency, with boards and CEOs pushing teams to discover possible use cases,” the report’s authors wrote. “Mid-to-large Providers are the exception, as they are early adopters with more resources to bring AI into production.” 

Plus, there are challenges to adopting AI at health care organizations. About half of providers and pharma executives pointed to security concerns as well as a lack of AI expertise at their organizations. The technology can also be costly to fully integrate, and finding AI-ready data can be a barrier too, particularly for pharma companies.  

However, adoption of AI scribes, which typically record a clinician’s conversation with a patient and draft documentation, has moved relatively quickly. Among providers, 30% have deployed a documentation assistant across their systems, while 22% reported the tools are being implemented and 40% said they’re currently piloting products.  

And though venture capital funding for startups touting AI tools has risen, many projects at health care organizations are being developed in-house and with large technology companies. Less than 15% are currently bought from startups, according to the survey.