Building Clinician Trust in AI: Leadership, Safety, Authenticity

Most narratives explaining why a healthcare AI program has been put on hold try to identify a technical problem. The model failed to deliver, the vendor promised too much, the integration was difficult, and so on. During a recent Signal Room conversation, Dr. Larry Kuhn referenced a less obvious and less easy-to-fix problem. Kuhn said that most organizations can keep pace with technological advancements and can check those boxes. The problematic part of the equation is people. They are the most valuable, most costly, and most problematic part of the organization, and they are overlooked by managers who are focused on tools.

Kuhn, who is a clinical psychologist and leadership consultant, is the founder of AspireView and the president of Prepare to Change. He has helped many organizations transition and evolve while preserving the human element. In his recent paper, Trust at Work: The Seven Dimensions of Credible Authenticity, he argues that trust is a measurable factor of a successful business, and that in a care-giving sector the trust tax slows everything down while the trust dividend speeds everything up. For healthcare managers who are introducing artificial intelligence to their clinical teams, these principles are highly relevant.

Trust Was Already Eroding Before AI Arrived

The setting Kuhn described is bleak and predates the algorithm. Using the examples he cited, over the past 50 years, trust in a fellow human has declined to an estimated 50% in the 1970s and 33% in 2025. Trust in the social structures people rely on has declined, a drastic fall in trust in governments, the police, ministers, doctors, and employers, who used to command trust in the 70% range and have fallen to the mid-20% range in the numbers he cited.

Organizations do not appear to be any better. He pointed out that only about one in five employees have high levels of trust in their employer, less than half of employees have trust in their manager, and only about one in four employees have trust in their CEO. Kuhn argued that this phenomenon was due to something more fundamental than workplace cynicism. In reference to Maslow, Herzberg, and Erikson, Kuhn argued that the first thing a human being does, even as an infant, is to assess whether the person holding them can be trusted not to drop them. As a clinician, he added a comment from his own clinical experience that a large number of people are unable to recall a caregiver who was consistently present to meet their needs, and this results in a pattern of insecurity in their subsequent working relationships. None of these numbers should be interpreted as conclusive proof, as they are the evidence he contributed to the discussion, but the trend is clear, and it means that leaders are not starting from a neutral position. They are starting from a lack.

Why AI Turns a Trust Deficit Into a Stall

Combine disruptive change with that lack, and you create the kind of dynamics that slowly erodes AI programs. Kuhn's metaphor was of a train leaving the station, and that motion was faster than traditional relationships and processes could keep pace with. When people feel something is threatening what they value, be it their role, their position, or even their self-worth, the reaction is an instinctive fear, and fear is the antithesis of openness.

He illustrated three patterns of responses to feeling threatened. First, the controlling response tends to be bullish. It manifests an increase in quick and assertive commitments, usually accompanied by a promise to achieve something almost impossible. Shutdowns can be categorized as moving away. In this case, responses usually result in cessation of participation or any offered opinion about the situation and disengagement. Finally, the false moving toward response means someone is attempting to mask identified problems under a false level of cooperation. The motives underlying all three responses reflect, according to Kuhn, two standards of evaluation. People are more strongly inclined to not appear bad than they are to look good. In an environment where fear of embarrassment is present, keeping silent is perceived to be the best choice.

This is problematic in the case of AI. The person who first detects a problematic situation within a cycle of work that is amenable to automation is likely to be the most apprehensive to even mention it, as doing so would be the de-facto closure of their own position. This is even more so in healthcare. Most patients do not arrive feeling relaxed or trusting, and there is a shortage of clinical positions. When people are on guard regarding the answer to the question, how would you like to change the work you do, they are likely to shut themselves down in order to remain safe. The most valuable knowledge to improve a program is the knowledge that is sequestered by fear.

The Leader's Job Is to Make It Safe to Raise a Hand

Kuhn's approach, in this case, offers a response to the silence. The first move is reliability. Do what you said you would do. He openly stated that everyone has a slip up every now and then, and said that a culture has developed in which good intentions are enough. However, people's trust is based on what they have witnessed a leader do, not what a leader promised. From the workforce's perspective, and this is central to Kuhn's point, you should underpromise and overdeliver, rather than the reverse.

The second move is authenticity, the willingness to show the positive and the negative. He told the story of an investor. He said that the fastest way for a founder to sink is to paint only the positive, because everyone knows there is a negative and they wait for it. The temptation, especially for a leader who does not want to look bad, is to reassure. However, the question everyone has focuses not on the positive, but on the negative: how does this affect me? And a leader's self-reassurance does not gain trust, it loses it. The absence of the hard truths shows that you do not trust your people, and they lose trust in return.

This is where the CEO's instinct to guarantee that no one will lose their job becomes a setback. Chris identified the dilemma: such a promise is one of the most dangerous a leader can make. This is due to the fact that disruptive changes genuinely reshape the work, and the IT transformations he has witnessed drastically reduce the workforce over a period of a few years. The honest reframing is not the guarantee of a workforce's safety, but a promise that they will be empowered to engage in work that is more fulfilling and meaningful, along with an acknowledgment that some aspects of the work will change, and not everything is predictable.

Relatability, Readiness, and Refusing the Sales Pitch

Kuhn's final moves all had the same effect: creating a space where people could express their opinions. Relatability means abandoning the pretense that leadership has all the answers. Saying I do not know exactly what is going to happen either, and I will keep you posted does more to foster trust than a rehearsed response that no one actually believes. When people feel they are being sold a final product, they conclude that there is no space for their questions and feedback, and as a result they take their concerns elsewhere and speak to each other about them outside the room where it matters.

In his account, readiness is the capacity to be present and accessible, which is exactly what most people said they never had. It requires active provision of space to listen to staff and to be proactive in creating the conditions for staff to articulate their fears, instead of waiting for them to emerge. The last phase is being rational: carry out the necessary research, ground yourself in the facts, and avoid the conjectures and the sales pressure, which staff can perceive. Kuhn's preference was to be transparent and direct: here is what we know, here is what we do not, here is what we are willing to commit to, without being a safety net for staff for every outcome.

His last point reframes the whole issue for any leader looking for the magic solution to the problem. Kuhn's position was that trust is a relational issue, and relational issues cannot be solved with more data, more online services, and more ChatGPT. There has to be a connection, and it takes vulnerability from both sides. This is not a quick fix achieved by holding a six-monthly town hall. This is extensive and sustained work that must be done before the conversation can include technology.

How Hutchins Approaches Clinician Trust in AI

We meet this condition constantly: a capable AI initiative that cannot get traction because the people expected to use it were never given a safe way to shape it. Our work treats trust and psychological safety as part of the deployment, not a communications afterthought — building the oversight, honest framing, and feedback channels that let clinical and operational staff raise concerns before a model reaches the bedside. That discipline is inseparable from responsible AI in healthcare and from the AI literacy that lets staff know when to question, override, or escalate an AI-supported decision. These themes run throughout The Signal Room podcast, where leaders working at the intersection of healthcare, AI, and human change describe what earning that trust actually takes.

Authoritative sources

• WHO — Ethics & governance of AI for health
• NIST AI Risk Management Framework
• Agency for Healthcare Research and Quality
• NEJM AI
• HIMSS