AI adoption depends on something most healthcare organisations don’t measure

The question may not simply be whether an organisation is ready for AI, but whether the organisation has behaved in ways that make people willing to trust it with AI.

---

Healthcare organisations are spending a great deal of time working out whether they are ready for artificial intelligence.

The assessments are becoming increasingly sophisticated. Boards want assurance about data quality, cybersecurity, privacy, clinical risk, model performance and accountability. Executives are looking at workforce capability, implementation maturity and the governance needed to oversee systems that may influence clinical and operational decisions.

All of this is necessary, but it may also be creating a false sense of preparedness.

An organisation can be technically ready for AI and still be profoundly unready for what introducing AI asks of its people.

The missing variable here is trust – not simply whether clinicians trust an algorithm, but whether people trust the organisation and the leadership that is asking them to use it.

When an AI system arrives in a hospital or health service, it doesn’t enter a neutral environment. It enters an organisation with a history.

Staff already know whether executives listen, whether consultation genuinely influences decisions, whether problems are acknowledged early and whether difficult information can travel upwards without consequence.

They also remember previous change, particularly the big transformations that may have failed – the system that was meant to save time but created more work; the restructure described as improving care that mostly felt like cost reduction; the consultation process that occurred after the important decisions had already been made; the concern that was technically welcomed but professionally costly to raise.

Those experiences don’t disappear when the next technology project begins. They become part of it.

This is one reason why we’re increasingly uncomfortable with the way healthcare uses the phrase “resistance to change”.

It’s a convenient diagnosis. When adoption is poor, the workforce is resistant. The response is then familiar: more communication, more education, more champions, more engagement.

Sometimes that is exactly what is needed, but often, this “resistance” is being misdiagnosed.

A clinician asking whether data generated by an AI-enabled documentation system could later be used to assess productivity is not necessarily resisting innovation. A nurse questioning whether an AI recommendation can genuinely be overridden is not necessarily afraid of technology. A manager wondering whether promised efficiency gains will eventually translate into fewer staff is not necessarily being negative.

They may simply understand how organisations behave.

That distinction matters because by introducing AI we’re asking employees to trust much more than the technology itself. We’re also asking them to trust that the purpose described today will remain the purpose tomorrow – that the limits placed around the use of data will hold; that the human oversight we assured will remain meaningful; that safety concerns will outweigh implementation momentum when necessary.

None of those assurances can be secured through a policy alone, because people decide whether to believe them largely on the basis of what the organisation has done before.

Imagine two health services implementing the same technology. They have comparable infrastructure, similar governance arrangements and equally capable project teams.

In one, staff have seen leaders acknowledge mistakes, alter decisions in response to frontline concerns and protect people who raise difficult issues.

In the other, consultation is largely performative, unsuccessful projects are quietly reframed as successes, and employees have learned that challenging senior decisions may be welcomed rhetorically but not always rewarded professionally.

On paper, both organisations may appear equally ready for AI.

They are not.

The second organisation has a deficit that will not appear on a digital maturity assessment. It has less institutional credibility available to support the implementation.

AI makes particularly heavy demands on that credibility because it touches areas that healthcare professionals care deeply about: judgement, autonomy, privacy, workload, accountability and, ultimately, patient care.

Leaders are therefore asking people to accept uncertainty around technologies whose future uses are not always entirely known.

That requires trust not only in current governance, but in future leadership behaviour.

And this also changes the questions boards should be asking.

Boards should still seek assurance about safety, explainability, cybersecurity, data governance and accountability. But they may need to spend more time examining the environment into which the technology is being introduced.

• Would employees feel safe challenging an AI-generated recommendation?
• Do people believe concerns would lead to investigation rather than defensiveness?
• If the technology produced disappointing results after substantial investment, would the organisation be willing to say so?
• Could a clinician credibly slow an implementation on safety grounds?
• And perhaps most importantly: would staff believe that they could?

A speaking-up policy is not the same thing as believing it is safe to speak. A governance committee is not the same thing as believing the committee will challenge a favoured project. A consultation process is not the same thing as believing the decision remains open.

And AI may make these differences much harder to ignore.

In fact, one of its more interesting effects on healthcare leadership may have relatively little to do with technology. It may expose organisations to themselves.

When an AI implementation encounters scepticism, leaders will be tempted to look outward: at workforce capability, professional conservatism, technological literacy or resistance to innovation.

Sometimes the more useful place to look will be inward.

• What has the organisation taught its people about what happens when they disagree?
• What have previous transformations taught them about whose interests are protected when things become difficult?
• What evidence have leaders given employees that governance will still matter when governance becomes inconvenient?

These are uncomfortable questions because they cannot be resolved by establishing another steering committee.

Nor can trust be manufactured at the start of an implementation.

It is built much earlier, through a series of fairly ordinary leadership decisions: whether bad news is welcomed, whether commitments are kept, whether dissent is tolerated and whether inconvenient evidence actually changes decisions.

That is why some of the organisations best positioned to benefit from AI will not necessarily be those that move first. They may be the organisations whose people have good reason to believe that if the technology creates a problem, somebody will listen.

There is of course a tendency to treat trust as the softer side of digital transformation: important, certainly, but secondary to architecture, governance and implementation.

We think that gets the relationship backwards.

In healthcare, trust may increasingly need to be understood as part of the infrastructure.

A technically sophisticated organisation with low institutional trust may be capable of deploying AI without ever becoming particularly good at using it. A high-trust organisation, by contrast, can have people question the technology without rejecting it, report problems without fearing that they will become the problem, and distinguish legitimate scrutiny from opposition to innovation.

That is a much stronger foundation for responsible adoption.

So, the question for healthcare leaders may not be simply whether their organisation is ready for AI, but whether the organisation has behaved, over time, in ways that make people willing to trust it with AI.

The technology may be new. But the judgement people make about whether to trust the leaders introducing it rarely is.

Dr Sidney Chandrasiri is the CEO of the Australian Institute of Health Executives.

Professor Luis Prado is the chief academic officer of the AIHE.

This article was first published by the AIHE. Read the original here.