Areas of tension
Despite the domain-specific differences, strikingly similar areas of tension recur in both sessions. They are described differently and experienced differently, but they structure the discussion on AI and professional practice in a similar way.
Efficiency versus depth
In both sessions, AI is recognised as a powerful catalyst for processes. Participants describe how repetitive tasks are completed more quickly and production capacity increases. At the same time, there is a concern in both groups that these efficiency gains come at the cost of depth and quality.
Within CMM, this translates into concerns about superficial creativity and predictable output. Within TI, the same tension emerges in the observation that developers use or accept code without fully understanding what that code does. As a result, debugging, maintenance and knowledge-building become more complex. Stijn Lambrechts (Cegeka) explicitly highlights this when he states that developers’ knowledge is at risk of becoming impoverished whilst that of AI is growing.
The underlying logic is similar in both domains: when speed is prioritised over understanding, the problem shifts to other levels within the process.
Use versus understanding
A second area of tension centres on the difference between using AI and understanding AI. In both sessions, a consensus emerges that AI becomes problematic when professionals accept output without insight into its origin, limitations and reliability.
Within CMM, this involves understanding creative choices and being able to critically assess generated content. Within TI, it revolves around understanding what code does and being able to explain why systems work or fail.
Both sessions thus implicitly formulate a similar redefinition of professional autonomy: not producing everything oneself, but understanding, assessing and being able to intervene when necessary.
Control versus dependence
Control forms an important normative framework in both sessions. AI may support, accelerate and generate, but participants consistently monitor the limits of that support.
Within CMM, this concerns creative control and consistency of content. Within TI, it revolves around technical control, system stability and traceability.
At the same time, both sessions make it clear that there is a tension between perceived control and actual control. Participants express confidence in their ability to steer AI, but concrete examples show how that control comes under pressure in complex situations or when under time pressure.
Responsibility remains with humans
One of the most consistent insights across both sessions is that responsibility and ownership remain with humans. In neither context is AI seen as an actor that legitimises decisions or assumes liability.
The professional remains responsible for the choices made, must be able to justify decisions and remains accountable to clients, users and stakeholders.
Professional autonomy is therefore not eliminated by AI, but shifts towards other – and in some cases more complex – forms of responsibility, in which the ability and willingness to exercise professional judgement remain central.