AI in healthcare: a tool, not a replacement

A new study shows that 357 Swedish doctors are using AI tools that aren't approved - simply because the approved ones don't exist yet.

That fills a real need, but it also creates risks for privacy, security and regulatory compliance. And it points to what I think is the decisive question: it isn't about whether we should use AI in healthcare - it's about how.

We trust technology every day. A plane takes off and lands with enormous safety thanks to advanced technology, and some cars can today drive you from A to B almost on their own. Yet we always want a human at the controls or behind the wheel - it's even required by law.

The exact same principle has to govern AI in healthcare.

A human - a specialist - must always review and decide when an AI tool has summarized or written something. The AI never works alone. It should never independently make a diagnosis, and never tilt a clinician toward a particular conclusion.

The AI's job is to free up time - so that the people who work with people get to work with people, not with documentation. The technology stays in the background. The human stays in the foreground.

AI should always be a tool that frees up time for relationships and better care - not one that takes over.

The study shows the demand is already there. The answer isn't to ban AI, but to offer approved tools with data protection, logging, clear governance - and human accountability for every decision.