In the next two to five years, health systems plan on prioritizing predictive analytics (46.8%), clinical decision support (43.6%), and diagnostic and treatment recommendations (39.4%). 

While many healthcare leaders are considering the near- and long-term uses of generative AI, they are also facing barriers to implementing the technology, including resource and cost constraints (48.9%), a lack of expertise (48.9%), and regulatory and legal considerations (35.1%). 

Ultimately, Berger noted that most health systems are still unsure how generative AI works, whether they should build or buy applications, and how to implement it safely. (Berger/Dries, Bain & Company brief, 8/7; Turner, Modern Healthcare, 8/8; Diaz, Becker's Health IT, 8/7)

The 2 things you should prioritize ahead of a generative AI strategy   

By John League, Vidal Seegobin, and Ty Alderhold

The easy accessibility and broad applications of generative AI have renewed interest among healthcare leaders. 

But here's the truth: Providers don't need a generative AI strategy. Instead, they need to figure out how AI can support and evolve their existing strategies. 

A plan to implement AI-based solutions is essential to drive the efficiency and savings we see leaders striving for in their technology investments. However, stakeholders that proceed without a strategic understanding of are putting the cart before the horse. 

It's not surprising that most provider organizations don't have generative AI strategies. The reality is that most of them can't afford to be first movers on generative AI. In terms of dollars, attention, and risk, the costs are either unknown or too high for many organizations. 

Cost pressures often make even obvious investments hard to justify. Leaders tell us that investments in digital health or AI that they could have made without much discussion (or maybe even without hesitation) just two years ago are now off the table. 

Additionally, pervasive and overwhelming demands on clinical and administrative staff often hinder efforts to master new workflows or submit to the requirements of change management work. 

In some ways, generative AI still seems inherently risky — and vendors aren't always as helpful as they could be. Provider organizations are skeptical of the prevailing message of "trust us" from many vendors about the quality of their models and the safeguards placed around hallucination and bias. Given the other costs, this opacity only further validates many providers' hands-off approach.

The Bain report also confirms what we know from our own conversations with health system leaders: they are curious about AI and they recognize its potential to reshape healthcare. At our Future of Hospital-Based Care roundtable earlier this year, health system leaders identified AI as one of the top three potential disruptors to their segment of the industry, behind only workforce challenges and new entrants to the market. 

However, provider organizations shouldn't spend their time building a standalone generative AI strategy. They should be doing two things to familiarize themselves with the technology and prepare themselves to deploy it to advance and adapt their existing strategies:

1.       Observe the market. We don't mean "ignore generative AI for now and think about it again in a year or two." Instead, we know that some provider organizations are actively identifying the teams and leaders who have interest in AI and giving them some latitude to evaluate what other organizations and vendors are doing. This includes both the emerging capabilities of generative AI and the ways that it is being adopted by clinicians, patients, and even other industries. Then, when the technology is cheaper and use cases are established, these organizations won't be starting from zero. 

2.       Survey your own operations. Many organizations probably already have some form of AI in use — even if they don't fully realize it. Clinicians and staff are using generative AI to support their work. Anyone with a smartphone can access ChatGPT, and many services and platforms will increasingly build in generative AI capabilities to their suites of services. We are finding that even some large provider organizations that are known for their use of AI do not have a complete understanding of how and where it is being used across their systems.