Conceptual definition of professional autonomy, independence and supervision

In this synthesis, professional autonomy is not understood as complete independence from AI, but as the ability of professionals to critically understand AI-integrated work processes, to manage them in a targeted manner, to assess their content, and to take responsibility for their outcomes. Autonomy therefore does not refer to avoiding AI, but to maintaining professional judgement within situations in which AI plays a structural role.

Within this analysis, autonomy refers to what professionals – and, by extension, students – must be able to understand, decide, carry out or account for with sufficient independence, with or without the support of GenAI. The concept thus aligns with the central project question regarding the desired level of autonomy.

Steering refers more specifically to the extent to which professionals actively guide AI systems during the work process. This involves formulating objectives, context, quality criteria and instructions, as well as the ability to iteratively adjust AI output when it does not align with the professional intention.

Within this analysis, professional dispositions refer to the way in which professionals recognise, activate and apply knowledge, skills and professional judgement within specific professional situations. In line with dispositional approaches, these encompass not only the ability to act appropriately (ability), but also the sensitivity to recognise relevant situations (sensitivity) and the willingness to act effectively (inclination).

The findings from both sessions suggest that professional practice in AI-integrated contexts does not depend solely on what professionals know or are capable of. Critically assessing AI output, taking responsibility for decisions, exercising control over processes and interpreting context meaningfully require not only knowledge and skills, but also a willingness to actively apply professional judgement when AI offers convincing, efficient or seemingly correct solutions.

The concept of disposition is therefore not used in this synthesis as a substitute for competences, autonomy or independence, but as a complementary interpretative lens that helps explain why professionals with comparable knowledge and skills nevertheless deal with AI-integrated situations differently and, in doing so, demonstrate different forms of professional practice.

The distinct concepts refer to different dimensions of professional practice, but are closely intertwined in practice. For example, a professional may possess in-depth domain knowledge but still only actively manage AI to a limited extent. Conversely, someone may use AI very actively or provide prompts without sufficient substantive understanding to assess the output professionally. Furthermore, professionals may possess comparable knowledge and skills, but differ in their sensitivity to intervene critically or their willingness to take responsibility when AI output falls short. It is precisely these tensions that form the basis for the scenario axes and typologies elaborated below.