Professional communities should recognise the core skills of their professions and protect them from AI.

Artificial intelligence has taken over the role of assistant in many professional fields. Doctors, researchers and software developers rely on it in their work. At worst, this results in a phenomenon described by researchers as deskilling: people’s professional skills deteriorate as they outsource their tasks to AI.

Academy Research Fellow in Theoretical Philosophy Kaisa Kärki discusses deskilling in a research article entitled Who is Responsible for Tackling Professional Deskilling?

“One of the problems with deskilling is that professionals have difficulty noticing it in their everyday lives, partly due to biases about AI reliance. When we converse with large language models and they respond convincingly, it easily leads to overreliance. However, these models also make mistakes.”

Kärki gives healthcare as an example: a doctor relying on a large language model without noticing its inaccuracies can make a medical error. Who is responsible for the error?

“Generally, philosophical theories of responsibility have suggested that an agent should be held responsible for those actions she has done with awareness. The inconspicuous biases associated with AI use pose something of a challenge to these viewpoints: after all, we don’t notice our own biases or, say, tendency to rely too much on AI. It’s therefore unclear whether we can be held responsible for the insidious deterioration of our skills and the mistakes that will result.”

Could we still manage without AI?

Kärki points out that, because deskilling easily goes unnoticed, individual professionals cannot be held responsible for related problems, at least not all of them. The issue is societal and structural.

Professional communities should identify those core skills of the profession that should be kept in human hands.

For example, pilots practise autopilot failure situations. Other professions should consider the same: how will people fare without AI when necessary?

“In medical education, it has already been argued that training in basic skills should include a stage where AI is not used. For instance, diagnostic reasoning is a skill central to medical practice, as is general understanding of the field and the associated tasks.”

The risk must be addressed now

If AI is increasingly used at the workplace without considering the problems that deskilling causes, we may end up incapable of training new professionals without AI. In other words, people could become too dependent on technology and AI, overlooking its errors.

The resilience of human communities is linked to people’s diverse skills.

“Deskilling involves major risks that humanity and professions should pre-emptively consider before skills decay at the community level,” Kärki says.

How to tackle deskilling

• Professionals must be made aware of the fact that AI use can insidiously cause deskilling.
• Even seasoned professionals should have their core skills tested regularly.
• Some tasks should be restricted to humans only, without the option of outsourcing them to AI.
• Workplaces should draw up AI use guidelines that consider skill decay.
• The higher education of professionals should include stages during which AI use is prohibited.
• Financial incentives should be developed to preserve skills.
• Legislation should be built to regulate AI use in fields where deskilling threatens processes critical to society.
• Professional communities should determine the core skills of their fields, which need to be kept in human hands.
• Skill decay caused by AI use should be studied more also in fields other than medicine.

Research article

The study received support from the Weisell Foundation, the University of Helsinki and the Research Council of Finland.

Kaisa Kärki: Who is Responsible for Tackling Professional Deskilling? 23 September 2026