Three experts on AI in the healthcare sector

The challenge for the modern supervisory or executive board member, however, lies not only in establishing policy, but in understanding the impact of AI on the human dimension and social solidarity. As algorithms increasingly guide the decisions for doctors and patients, the role of oversight is shifting toward ensuring transparency, safety, and ethical frameworks. It is no longer sufficient to delegate AI to the IT department; effective leadership requires that the organization’s top itself understands how this ‘black box’ functions, identifies opportunities, and recognizes the risks associated with relying blindly on automated outcomes. In the dialogue below, leaders from the healthcare sector, the insurance industry, and government explore how they apply AI in practice and where the limits of technological progress lie. Moderated by Yoen van der Linden, managing consultant at SeederDeBoer, Stephanie Klein Nagelvoort-Schuit (vice chair of the executive board of the UMCG), Annemarel de Vos (Director of Health Insurance at a.s.r.), and Mark Janssen (chair of the executive Board of the Dutch Healthcare Institute) discuss how they deploy AI in practice and where the limits of technological progress lie.

Let us get straight to the point. AI is everywhere, but what does it actually mean for your own daily professional practice? Stephanie, you are responsible for a massive organization within the UMCG. How AI-ready are you yourself?
Stephanie Klein Nagelvoort-Schuit: ‘I firmly believe that as a leader, you cannot provide direction if you do not understand the technology yourself. That is why, at the UMCG, we on the board of directors are piloting virtual assistants ourselves for decision preparation and scenario analysis. In my personal life, I have even built my own Karpathy Second Brain. Being actively engaged with AI helps me see the opportunities and risks coming our way in healthcare – we need to prepare for them today. We are also seeing a gap emerge between the frontrunners and the rest of the organization; we must prevent people from viewing AI as an oracle rather than as a second brain with which you engage in dialogue.’
Annemarel de Vos: ‘At a.s.r., we see a similar trend. Our new CEO, Ingrid de Swart (see: ‘This job is even more intense than I thought’ on page XX of this magazine), views AI as a key strategic tool, and that translates into a clear tone at the top. We have set up a ‘incubator’ within the healthcare business where we experiment with AI concepts that we can then scale up more broadly within a.s.r.. For example, within health insurance, we have deployed AI for the processing of foreign claims to take a large part of the investigative work off the hands of employees and support them with the analysis. But crucial to this is that we have established an AI Quality Board where legal experts, ethicists, and risk, compliance, and cybersecurity experts assess applications against laws and regulations as well as potential risks. This allows us to learn faster as an organization and implement solutions more quickly – but in a controlled manner.’
Mark Janssen: ‘At the Dutch Healthcare Institute, we naturally also examine our own processes. We have to evaluate massive dossiers on new medications, and we are investigating whether AI can help us speed up those assessments by intelligently summarizing research results. But my concern also extends to a broader level: how do we use AI to identify inappropriate care? We have seen in pilots that algorithms can help us uncover patterns where medications are prescribed that have long since been removed from the guidelines. That is where AI truly adds societal value to the accessibility and affordability of healthcare. But AI also raises fundamental questions: can patients trust that their data will be used responsibly? How do we prevent algorithms from exacerbating existing health disparities? And how do we ensure that healthcare professionals are supported by AI without relinquishing their own professional judgment?’

That sounds like a significant efficiency boost. The Cabinet is even aiming to halve the administrative burden through AI. But Stephanie, you just mentioned the risk of the 'oracle'. How do we prevent healthcare professionals from blindly relying on an algorithm?
Stephanie Klein Nagelvoort-Schuit: ‘That is the essence of good governance in this transition. You have to retain core skills. Just as children at school still have to learn their multiplication tables despite the calculator, our doctors and nurses must retain their clinical judgment to recognize when AI draws an unlikely conclusion. At the UMCG, we learn that successful teams invest in different types of employees: the innovators who can build things themselves, the gamechangers who can transform processes, the vast majority of users, and the group that needs more time. Only by recognizing these different roles can you responsibly integrate AI into the line organization.’
Annemarel de Vos: ‘Trust is the key word here. If employees or customers view the outcome of an algorithm as a black box, it leads to resistance or conversely, dangerous complacency. We believe human intervention is essential; AI can support employees in preparing, organizing, or flagging cases. Where assessment is necessary, human control and responsibility remain safeguarded, especially when an outcome has major consequences for an insured person or healthcare provider. When we do it right, AI is not visible to insured individuals as technology, but rather feels like less hassle and faster, more transparent assistance.’

Mark, you have a responsibility for the system as a whole. When it comes to that trust, do you also see a role for AI in improving collaboration between hospitals and insurers? Because that is where things sometimes get tricky.
Mark Janssen: ‘Unfortunately, the healthcare system is still often structured around mistrust and control. Doctors verify everything themselves, and insurers check for compliance. AI can drive a breakthrough here, provided we get the data infrastructure in order. If we work with the same standards and definitions, a ‘single truth’ emerges that eliminates the need for debate. Then we can focus on discussing why there are variations in practice across regions, rather than questioning each other’s data. But to do that, we must stop creating our own silos and pilot projects that never extend beyond their immediate region.’
Stephanie Klein Nagelvoort-Schuit: ‘I recognize that mistrust. For example, we have a program for appropriate care for elderly cancer patients, in which a quarter of them, after a thorough discussion, opt for a different treatment plan – usually a less invasive one. This results in the same life expectancy, while these patients experience fewer complications and spend less time in the hospital. Yet we received questions from the insurer: ‘Are those patients not just going to the neighbors?’ To a doctor, that feels like a lack of understanding of the medical substance. AI and better data sharing can provide an objective view of what is actually happening here, so that we can improve the process instead of monitoring each other.’
Annemarel de Vos: ‘And that is precisely the opportunity for us as insurers. We have a lot of data from our statutory mandate. We can help hospitals by demonstrating that their approach does indeed lead to better outcomes without patients seeking care elsewhere. But that does require us to make agreements upfront about who invests and who reaps the benefits. If investments in, for example, digital applications demonstrably contribute to appropriate care, accessibility, or quality, we must also consider how the benefits and investment costs are distributed equitably within the care chain.’

A theme I am hearing more and more often is the tension between the far-reaching personalization enabled by AI and the solidarity of our healthcare system. Annemarel, how does an insurer view this?
Annemarel de Vos: ‘That is a fundamental risk. The better we can predict outcomes, the more consciously we must protect solidarity. You want to prevent AI from being used in a way that leads to risk selection or unequal access to care and insurance. For a.s.r. solidarity is the starting point: technology must not undermine the foundation of the system. Earlier detection, better referrals, less administrative friction, more self-service, and greater transparency. These are all things where, in my opinion, AI can make a fundamental contribution without compromising solidarity or public values.’
Mark Janssen: ‘I see a tension there, but also a solution, in the legislation surrounding appropriate care. Personalization means we provide care that suits you; for one patient, that might involve an intensive treatment process, while for another, it might mean stopping after the diagnosis because that person has different needs. That is both personal and based on solidarity, because we collectively decide to pay for what truly adds value for the individual. The insurer has a duty of care in this regard to make that customized approach possible.’
Stephanie Klein Nagelvoort-Schuit: ‘Exactly. AI can help us determine that ‘suitability’ much more quickly based on vast amounts of scientific literature and patient data. But the ultimate moral choice – what do we consider an extra year of life to be worth, and which side effects are acceptable –remains a human, societal consideration.’

Looking at the role of the supervisory board and the executive board, what is your most important advice for good governance on this issue over the next two years?
Stephanie Klein Nagelvoort-Schuit: ‘Invest in leadership that understands it. You cannot oversee something you view as a black box. Make sure there are people at the top who know what AI does and where the data comes from. And dare to converge: stop the fragmentation and work together nationwide on standards.’
Annemarel de Vos: ‘Do not be afraid to experiment, but do so within clear parameters. Make AI a strategic priority and ensure that the entire company is on board. Successful organizations will eventually not distinguish themselves by their access to technology, but by how they connect that technology to trust, service, and social value.’
Mark Janssen: ‘The Netherlands does not so much face an AI challenge as it does a data challenge. My advice is therefore: focus on the data. Without reliable data, good information standards, and secure data exchange, the impact of AI will remain limited. As stakeholders in the healthcare system, we must take the lead on data availability. Only then can we use AI for what it is truly meant to do: keep healthcare accessible, affordable, and of high quality for everyone.’

What are your expectations for the future of AI in healthcare over the next five to ten years?
Annemarel de Vos: ‘I expect that in ten years, AI will be the invisible engine behind many healthcare processes – not as a replacement for professionals, but as support for administration, decision-making, and organizing appropriate care. This will allow scarce capacity and resources to be deployed more effectively. For a health insurer, the greatest value lies not in the technology itself, but in the ability to keep care accessible, affordable, and personalized in a time of increasing demand and staffing shortages. At the same time, public values ​​such as solidarity, privacy, and transparency will become even more important. Precisely because AI is becoming increasingly powerful, we must focus more consciously on building trust. Ultimately, success is not determined by how much AI we use, but by the societal value we create with it.’
Mark Janssen: 'I also believe that AI will move in this direction and that the focus should be on value, not technology. For me, the discussion about AI ultimately revolves around trust. Trust from patients that their interests come first. Trust from healthcare professionals that technology supports them. And trust from society that innovation contributes to public values.’

This article was originally published in Management Scope 07 2026.