AI is losing the trust battle. What does that mean for healthcare?

AI was always going to be a bubble. The only question was what was going to pop it – and increasingly, it looks like the answer is going to be trust. 

AI has been making big, rapid, sweeping, moves into everyday life. Frog-marched forward by a new crop of Big Tech barons, it has been dramatically overhauling everything from the way we consume content to the shape of the communities that host resource-intensive data centers. 

But for the everyday consumer, it might be moving too far too fast, especially since we seem to have largely skipped past the part where we lay down meaningful, socially responsible ground rules for when, where, and how we integrate AI-generated insights into critical activities, from the classroom to the clinic. 

The result is a growing gulf between how much we use AI technology and how much we trust it.  

And when lives are on the line, that trust gap has major implications for when AI can become a high-value utility – and when it turns into an unacceptable risk. 

With healthcare organizations investing billions of dollars in AI with the goal of streamlining workflows and improving consumer experiences, where does this disconnect leave us? And what can turn the tide so that AI can demonstrate its real value at the most appropriate points in the care delivery process?  

Inching toward a failing grade on accuracy and experiences  

In many ways, AI is simply not meeting the expectations of its users. Recent surveys have shown a sharp downturn in optimism and enthusiasm around AI technologies, even as consumers are being nudged into using these tools more often and in more places in their lives. 

Even the most engaged AI users are less positive about their tools than they were a year ago, a new Gallup poll shows, with particularly pronounced shifts among GenZ users, who are currently 14 to 29 years old. 

About half of GenZ respondents use generative AI daily or weekly, yet strong agreement or agreement that they feel “excited” about AI has dipped by 14 percentage points to just 22%, while “hopefulness” has fallen nine points to 18%. Meanwhile, anger about AI technology has increased nine points to 31% and anxiety about AI has stayed steady at 42%.  

GenZ represents the next generation of decision-makers in healthcare. They are students or early-career care providers who will be using these tools in practice. They are becoming young parents with children to care for.  And it won’t be long before they will be assisting aging family members with their complex needs.   

The fact that this demographic is cooling so quickly on a category of technology that has become so ubiquitous in their lives raises red flags about how well the implementation process is going. 

The sentiment shift isn’t limited to GenZ.  Public openness to AI is declining across generations, with a separate survey from The Ohio State University Wexner Medical Center finding that support for AI has dropped from 52% in 2024 to just 42% in 2026.  

Pew Research adds that users of AI chatbots for health concerns are more likely to find the tools convenient (48%) rather than accurate (18%), which isn’t too surprising given the flurry of studies indicating that the real-world accuracy of tools like ChatGPT and Claude is often variable at best and dangerous at worst. 

Clinicians are expressing concerns that mirror those of their patients.  Close to two-thirds (63%) of physicians in an AMA survey are still more concerned than excited about AI in clinical practice, although the vast majority are not worried that AI will actively increase their stress or burnout. 

Instead, providers are more worried about “deskilling” across the professional community, with 88% of respondents to the AMA survey at least moderately worried about skill loss in students, colleagues, and themselves.  Some physicians even rate their genAI-using peers as less competent than those who rely only on their own skills, highlighting that a personal lack of trust in technology can spread negativity and divisiveness through an entire practice if left unchecked.  

Building trust with intention and intelligence

Healthcare organizations cannot afford for trust to be the limiting factor in the trajectory of their AI investment. 

It’s still early enough in the process to shift the sentiments of both consumers and providers – if organizational leaders prioritize operational and structural improvements that matter most for moving in the right direction. 

Lean into user-controlled transparency as a baseline requirement

Consumers and clinicians have stated again and again that transparency is a fundamental requirement for earning their trust in AI.  But transparency has to be more than just labeling when an AI tool was involved in a decision (although that’s a valuable step to take). 

True transparency gives the user the ability to look under the hood and how an AI tool arrived at an answer, including what data was used, what inferences were made, and how confident the model is that it has arrived at the correct answer. 

For example, a study from Mayo Clinic found surfacing confidence scores to clinicians is an effective way to boost trust in high-value answers while also helping clinicians retain control over their decision-making when the outputs are less certain. 

Clinicians overrode 73.9% of outputs when transparency was minimal, compared to 49.3% with moderate transparency. With the implementation of a confidence-calibrated framework, overall override rates fell to 33.29%. High-confidence predictions (90–99% certainty) were overridden just 1.7% of the time, while low-confidence predictions (70–79% certainty) were rejected at a rate of 99.3%.  

Similar strategies can help address several of the top concerns about AI in clinical decision-making by ensuring clinicians are always steering diagnosis and treatment pathways and reassuring patients that even when AI is involved in their care, humans are always in the loop and aren’t blindly trusting an algorithm. 

Deploy AI intelligently to reduce friction instead of creating it

Trust is built and reinforced in moments where technology makes something easier, faster, or safer, not where it adds another layer of complexity to a process.  AI is only going to meet that brief sometimes. 

Exhibiting judicious restraint and embracing an honest perspective about AI’s strengths and weaknesses will be crucial skills for healthcare leaders, especially when they are being constantly bombarded by pitches about the latest and greatest tools to immediately solve every possible problem. 

Organizations that are selective and strategic about deploying AI – and those who are willing to pivot quickly when something isn’t achieving the desired results – will be more likely to generate organic, sustained trust within the user base.   

A careful and thorough evaluation of the user journey, clinical or consumer, with input from every individual who touches that workflow, is an essential first step for deciding if an AI tool is an appropriate fit for the use case. 

Make AI governance robust, visible, and responsive

Safety, reliability, accountability are non-negotiable in the healthcare industry, and trust in these fundamentals isn’t built on vague promises that the organization has guidance in place. 

When deployed correctly, governance can be a competitive differentiator that proves an organization is worthy of user trust.   

It’s not enough to “implement a strong governance framework.”  The key is demonstrating that when an AI system fails (and one of them undoubtedly will), the issue is identified quickly, communicated clearly, addressed with urgency, and translated into a lesson that actually changes how the organization approaches issues in the future. 

When users can see that governance is visibly working and accountability is real, they are more likely to feel comfortable engaging with AI tools, and may even feel more confident exploring the possibilities further when they know there’s a reliable safety net to support them. 

Aligning confidence with capabilities to keep AI moving forward

The level of trust in healthcare AI will determine whether the bubble pops all at once or if it simply deflates to a more sustainable size. 

Minimizing the trust gap will take deliberate, visible effort to improve transparency, roll out AI judiciously, empower users to actively engage with AI decision-making, and create structured accountability frameworks that effectively nurture AI’s promises to make healthcare safer, more accessible, and more effective. 

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].