Table 6 Data extracts for sub-theme: healthcare is unique

And I think in healthcare is a one level higher concerning. So, what level of security you want to have? Because you don't, you say “okay, if the autonomous driver in a taxi is stopping at the wrong place, or whatever. It's annoying but I don't care. If it makes an accident, that would be bad but if I'm protected, okay”. But if I go to healthcare professional, and I have a disease and I have a health problem. I will, I'm looking for the maximum security that this, that they, that I'm recommended the right therapy, or I got the right consultation and this is what people do all the time. Because this is where they go to one physician, and they don't go to another, or they fly around the world to go to the right surgery blah-blah-blah. So, I think, we thought a lot about how, in which position would we see AI in the treatment concept for patients... [Rn1 (MEAI)]

You have to think very carefully about the used context of the particular AI system you are implementing and how it's going to fit institutionally. How it's going to affect workflows and so on. Because we have already talked about, a lot of the ethical issues, sort of, there are risks that could happen if you implement it. So, even a good AI system could be implemented in ways which means that it creates problems in the healthcare professional-patient relationship, or other problems. [Rn5 (BE)]

Healthcare is people’s health outcomes, it’s not just something you can iterate, test, and count on that it is something that people want. Even if people want this thing, it does not mean that this is what they should get, right? So, the role of experts in healthcare is clear that, it can’t go without it, because we can see what can happen with marketing especially, right? Ways that you think a product can save your live or whatever, and if it is not actually doing what is told. We live in a world of marketing and narratives. And so, I think in healthcare it’s really important that those things are accurate. [Rn25 (DH)]

If we keep stating that the healthcare is, it's different to retail, for example, then different type of decisions or to guide the way that we engage private companies and the framework in which these private companies are operating. So we cannot have the same framework for data use for the development of AI algorithms to, you know, to use in retail and the same type of AI, like regulations for AI tools to be used in cancer diagnosis. So, these are very two, very different activities. Even if from the computational problem, they might seem very similar, in terms of the context in which they operate and, you know, and the aim is, it's a thing, I would say are fundamentally different. [Rn36 (BE)]