In discussions about AI, one temptation surfaces in almost every leadership team: the desire to control the applications. Ironically, this is often what causes initiatives to lose momentum.
Executive teams play a critical role in the AI transition but not in the way many assume.
Leadership’s most important task is not selecting tools, designing workflows, or deciding exactly how AI should be used in daily operations, says Carl Heath.
– Those decisions are usually too far removed from the actual work. When leadership tries to make them, decisions become slow, generic, and quickly outdated.
Strategic choices
What leadership must own are the strategic choices. This includes formulating the principles governing AI use, defining risk appetite and ethical boundaries, and clarifying where automated decision‑making is acceptable, and where it is not.
– Leadership also owns the resource decisions: how much time is allocated to learning and experimentation, and how clearly that time is protected from short‑term delivery pressures.
Creating real mandate
One frequently underestimated task is creating real mandate. Saying “teams are free to experiment” is not enough, Carl Hetah states.
– Mandate requires legitimacy. Who is allowed to make which decisions? What happens when something doesn’t work? When organizations lack clear answers, people either hesitate or bypass the system entirely.
Avoid decision paralysis
Problems arise when leadership tries to retain control over applications. The result is often decision paralysis, fear of getting things wrong, and eventually a culture where employees either wait for guidance that never comes or find their own way regardless.
The main exception concerns IT security, information security, and regulatory compliance. In these areas, leadership should engage more directly. Even then, the role is to set boundaries, not to micromanage implementation.
– In AI, destructive leadership is rarely about over‑ambition. More often, it is the attempt to reduce uncertainty through control. When what is truly needed is trust within clear boundaries, says Carl Heath.
Three questions for the executive team:
- Which decisions can only we make — and which should we let go?
- Have we created real mandate, or merely stated intent?
- Are we trying to eliminate uncertainty — or build the capability to manage it?
Carl Heath is a senior researcher at RISE and a doctoral researcher at the University of Gothenburg, focusing on leadership, AI and digital resilience. His work explores how organisations can navigate technological change while strengthening human judgement, trust and agency. He approaches AI not as a technical issue, but as a leadership and organisational challenge rooted in learning and everyday practice.