Is AI really ready to save a healthcare workforce in crisis?

It’s not much of a newsflash to say that it’s pretty grim out there for healthcare workers, both on the front lines and in the back office.  Hopes for a speedy recovery of the workforce after COVID have long since faded, and ongoing economic turmoil has only made it more difficult for employers and employees to get on the same page in terms of compensation and expectations. 

Shortages of qualified workers aren’t just making it more difficult to come home from a shift with some gas left in the tank: it’s also affecting patients and how they experience care. Widespread shortages are leading to noticeable delays in access to services, less time for patients to work with their providers on important issues, and rising concerns about safety across the clinical environment. 

So it’s AI to the rescue, right? Headlines are abuzz with tales of robot nurses helping out their human colleagues, while EHR companies boast about ambient technologies and their game-changing AI agents that will make it faster and easier for overworked clinicians to knock out administrative tasks. 

And while there’s some increasingly impressive evidence that AI tools truly can make a difference in clinical care and administrative efficiency, an important question is emerging: will the ROI be strong enough – and scalable/repeatable enough – to make a measurable dent in the devastation that the deepening workforce crisis is poised to unleash? 

At the moment, the industry is hoping for “yes,” but reality might not be quite matching up with the expectation. Case studies showcasing success are promising, but are often limited in scope, while industry reports show mixed opinions about trust in AI and confidence about how leaders are infusing new technologies into human-centered workflows.  

There’s no question, however, that AI has the potential to be a powerful tool to jolt the healthcare industry out of its workforce woes. The challenge is understanding where, why, and how to use it most appropriately to compensate for – or better yet, get one step ahead of – the workforce’s seemingly inevitable breaking point.  

Quantifying the obvious: just how bad is the workforce crisis?

 Executive leaders and HR departments across the country (hopefully) don’t need poll numbers to recognize that workers aren’t very happy with the way their careers are going. But for the rest of us, a new survey from The Harris Poll and Strategic Education can put some context around the depth of the dissatisfaction. Here are some highlights: 

• More than half (55%) of employees say they’re going to be looking for a new job in the next year, and 40% of those are going to be searching the listings at other organizations.  
• Only 1 in 5 employees feel like their employers are invested in their long-term success, and less than 1 in 3 feel very valued by their employer, which is leading to a lack of loyalty to their organizations. 
• A staggering 80% feel like their employers take them for granted, with only 37% of workers expressing high levels of satisfaction in their current role. 
• On-the-job burnout, lack of adequate compensation, and limited advancement opportunities are among the top reasons why workers are searching elsewhere for employment opportunities, with younger Gen Z workers and millennials (those with potentially decades of service left in their careers) among the most likely to exit the industry early. 

With clinician shortages projected to top 1 million by 2037, including half a million nurses, this is bad news for everyone, particularly patients.  Research has shown that clinicians unhappy with their work environments and/or management teams are more likely to make errors and less likely to voluntarily speak up about problems, which dramatically increases the risks of patient safety events, care delays, and preventable mortality. 

AI tools for workforce woes: sham or solution?

Healthcare organizations have rarely adopted anything as quickly as they’ve jumped on the AI trend. In what seems like a matter of months, close to 90% of organizations have put some type of AI tool in place, including generative AI capabilities (genAI) to support clinical, administrative, or operational use cases.  

Some of these tools, such as ambient listening technologies that assist with generating clinical documentation, have rapidly seen resounding success.  

At Carle Health in Illinois, for example, ambient listening tools from Nabla have resulted in more than half (55%) of clinicians saving at least one hour a day on documentation, and close to 80% of users expressing a desire to transition fully away from traditional documentation methods to using the new technology, said David Lovinger, MD, Chief Informatics Officer at Carle Health, during a recent CHIME event. 

“I’ve had some of our older doctors say they would fight me if I tried to take this away from them,” he said. “I’m not going to take that risk. It’s been such a successful project on so many levels.”   

But other AI experiments haven’t been so fortunate. In fact, generative AI pilots actually have a fairly dismal track record of success, according to a recent report from MIT, with up to 95% of organizations across multiple industries failing to return measurable ROI. 

Overall, the MIT report identified healthcare as one of the industries least disrupted by genAI so far, with success largely limited to the aforementioned documentation tools. Perhaps it’s no wonder, then, that healthcare leaders in the survey also reported few plans to reduce hiring based on the impact of genAI pilots thus far. 

Instead, they may be looking to AI as a tool to help them increase their hiring rate simply to compensate for the attrition created by poor job satisfaction. By reducing avoidable costs and accelerating revenue recovery through automation, organizations could theoretically free up additional funding to bring in flesh-and-blood workers to perform the clinical and administrative tasks that only humans can do – and do them at the highest level possible, which could increase engagement and satisfaction with the daily grind. 

But there could be some bumps in the road there, too. In some cases, AI might be taking over the wrong tasks, leading to mistrust in new technologies and potential negative impacts on patient care.  A 2024 survey from National Nurses United shows deep concern among nursing staff that AI is already intruding on their clinical judgement, and may be putting patients at risk. 

Half of the 2300 respondents to the poll said that their employers are using AI tools based on EHR data to determine patient acuity and the need for nursing care. Among those, 69% said the algorithm’s results differed from their own assessments.  

Other nurses said they’re not able to change AI-generated risk scores or needs assessments, limiting their ability to use their clinical judgement to advocate for their patients, and a whopping 60% said they simply don’t trust their employers to keep patient safety top of mind with algorithms in the clinical environment. 

“While our employers argue AI will help us, they’re using these technologies to erode our ability to practice our clinical judgment,” said Aretha Morgan, an RN in emergency pediatrics at New York Presbyterian and a board member of the New York State Nurses Association. ”Healing our patients requires time that we are not given by our employers and so AI is used to justify the crisis conditions they’ve made us work under.” 

For these nurses, at least, the assertion that AI can positively transform their job descriptions doesn’t hold water. Instead, AI is already becoming intrusive and frustrating, with its adoption driven more by high-level business pressures than a desire to improve patient care – exactly the same sort of issues that are leading workers to say their employers don’t value their input or needs. 

Hitting the sweet spot with intelligent, involved implementation

Is there a way to make AI actually work for everyone from the CFO to the CNA to keep the workforce in good trim?  There certainly is, and it involves some of healthcare’s other favorite buzzwords: strategic implementation, collaboration, and a non-punitive culture of safety around feedback. 

Selecting the right tools for the right job

AI is everywhere, but that doesn’t mean it’s uniformly mature or appropriate in all situations. With 77% of respondents to a late 2024 survey published in JAMIA saying that a lack of maturity is their number one barrier to success, and cohorts like the National Nurses United respondents concerned about where AI is being inserted into the workflow, leaders need to have a strong understanding of what’s really going to make a positive impact in their high-priority areas. 

According to the JAMIA survey (and plenty of other anecdotal evidence), applying AI to clinical documentation tasks seems to be returning the most value for both the organization and the workforce. More than half (53%) of respondents to that survey say they’ve seen significant success with AI-enabled documentation tools, compared to a surprisingly low 26% who have achieved success with revenue cycle management tools, and just 9% who have implemented patient-facing chatbots and other apps to engage patients in care. 

The MIT report reinforces this by pointing out that the most successful genAI adopters start by embedding AI into non-critical processes that can quickly demonstrate clear value, then scale systematically into core workflows – a clue to why ambient listening is a big hit, but advanced clinical decision support is drawing skepticism from users.  

Upskilling the workforce to take an active role in implementation

Healthcare employees are desperate for opportunities to grow and learn, the Harris Poll revealed, with a third of respondents recognizing that continued education will be crucial for advancing their careers.  

And while 43% of employees say that AI-related skills will be the most important area for their education over the next few years, 60% of employers admit that it’s difficult for them to provide AI skills and training on the job for their workforce. That’s leaving 1 in 3 healthcare workers feeling like they’re behind their peers in other industries, and fueling dissatisfaction with how their employers manage their long-term career development. 

Creating opportunities for upskilling, and giving AI-minded staff members the resources to actively participate in decision-making, is critical for seeing broad adoption that authentically enhances the workforce experience. 

For example, Mayo Clinic has announced a new AI tool created by their nurses for their nurses that displays curated, nurse-specific information on a single page within the EHR, complete with direct links to evidence-based practice resources. 

“To see a concept I was passionate about, AI-enhanced communication, actually get built — and to be invited to help shape it — reinforces that frontline nurses’ voices matter and that we have the power to influence the future of care,” says Brendon Bloomfield, a registered nurse in Psychiatric Acute Care at Mayo Clinic Hospital Rochester. 

Nurses can even directly submit feedback on what’s being called a “minimum lovable product” to shape its evolution and continue to drive improvements, which ties directly into the last main point: developing an organizational culture of valuing workforce input and creating a safe environment to report concerns. 

Developing an authentic culture of collaboration for the workforce

Investment in employees takes many different forms, from offering tuition assistance or other opportunities for career advancement to simply making it safe to voice concerns if digital tools aren’t doing exactly what they’ve promised. 

As the great AI experiment continues, and reaches into more and more areas of the workflow, healthcare leaders are going to have to rely on end-users to keep tabs on the hundreds or thousands of new algorithms, because they just won’t have the dedicated IT resources to identify every risk that might arise, especially in rare but critical edge cases that touch patient care. 

Fortunately, for the most part, executive sponsorship hasn’t been a problem with AI so far.  In fact, it’s the C-suite that seems to be pushing hardest to bring AI into the ecosystem. But with top-down pressure comes a reluctance for workers to push back against their bosses, and an existing baseline of demoralization in the workforce can make responsible whistleblowing even more unlikely.  

AI development must be a two-way street, especially with such an increasingly fragile bond between employees and their employers. To safeguard patients and keep the workforce as intact as possible, leaders cannot afford to limit AI implementation to an IT project.  

Instead, they must recognize that their employees are feeling desperate for a sense of value and belonging in their organizations, and are simultaneously itching for opportunities to bulk up their resumes with the latest technical skills.  

It’s a recipe for success if leadership is willing to embrace it. Not only can AI tools save time and reduce burdens within the everyday workflow, but it can be a lever for long-term employee loyalty and workforce sustainability. 

Organizations that take a visionary approach to holistic collaboration around AI adoption may be more likely to see workforce stability despite the challenging conditions facing the industry. It’s not the only solution to the workforce woes, of course – there is still a fundamental shortage of humans to contend with, and the answers to that problems are bigger than technology adoption alone – but it could be a valuable strategy for keeping healthcare humming along as smoothly as possible for the present.  

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].