Home›Leadership Insights›What leadership must be able to do when machines think along
Artificial intelligence
Nine capabilities, and the one question no machine can answer
As leaders, we face the task not only of understanding new technologies but of adapting our own role to changing demands. The question is: what capabilities does it take to lead in a world shaped by artificial intelligence?
The following nine capabilities emerged from the work of the Future of Leadership Initiative, from our own practice and from the discussions at the MLI Leadership Summit. Each is backed by an example, because a list of capabilities without examples is worthless.
In the age of AI, it is easy to get lost in technical possibilities. The potential only unfolds when the technology serves a purpose that can be named.
Example: A team that uses AI to improve the customer experience must not look only at the technical questions. The focus has to be on what customers experience differently afterward. That orientation gives the technology a direction.
Innovation in AI calls for an exploratory stance: thinking in cycles, building rapid prototypes, learning from successes and from mistakes alike.
Example: Instead of waiting for the perfect solution, start early and gather feedback continuously. That way, whatever does not hold up can be adjusted in time.
For all the technology, people remain at the center. The ability to bring people together and strengthen relationships decides how well disciplines work together.
Example: Teams working on AI develop tensions between technical and business experts. As a leader, you create the space in which both sides understand each other — usually by forcing the dialogue rather than hoping for it.
Leadership must recognize early what opportunities and risks AI holds for its own industry — and act before pressure forces it to.
Example: If your industry is automating more and more, start upskilling while you can still choose whom to upskill.
In a world that keeps changing, reliability and trust are the scarce resource. Leadership has to keep its promises, even when that becomes uncomfortable later.
Example: If you assure your team that introducing AI will not cost jobs, you have to deliver: through retraining and by integrating people into new roles. If you cannot deliver, it is better not to promise.
Decisiveness means taking responsibility in uncertain situations and deciding even though not all the information is available. It never will be.
Example: Moving an AI application into production usually means deciding against resistance, and bearing the consequences.
The development of AI is marked by uncertainty and ambiguity. Leadership has to remain capable of acting without clear-cut answers.
Example: When a solution does not deliver the expected success, the reaction decides whether the team will dare to try something again next time.
Introducing AI raises questions that cannot be answered technically. Data protection, traceability and the question of what data will not be used for belong at the leadership level.
Example: Your company could save considerable costs with AI, but at the expense of data security. Deliberately deciding against it is a leadership decision, not a legal question.
AI brings a complexity that cannot be delegated. A basic understanding of its logic and applicability is indispensable.
Example: Instead of relying on expert opinion, build up a basic knowledge yourself. Only then can you judge whether a recommendation serves your initiative or the vendor.
Nine capabilities are nine too many if you try to develop them all at once. In programs we work on one, rarely two. The rest are deliberately left aside. A list of capabilities is a map, not a work plan.
And an observation from our own practice: AI has reduced our team from ten people to three. That was not an efficiency measure but the result of analyses, drafts and summaries no longer tying up human working time. What remains is the work that happens between people, and the question of what for.
AI does not relieve us of leadership. It relieves us of our excuses.
The nine competencies are a snapshot in a field that changes within months. What counts as judgment in dealing with language models today may look different in two years. The list is therefore a frame for thinking, not an examination catalog.
It does not replace technical expertise. Leadership does not have to be able to build AI, but it needs people in the organization who can, and it has to know when to believe them. Without these specialists, any list of competencies stays theory.
And not every competency rests with the individual leader. Whether experiments are allowed, whether mistakes are discussed, whether data may be shared is decided by the organization. Anyone who reads the nine points solely as development goals for individuals overburdens the individual and lets the structure off the hook.
© MLI Leadership Institut GmbH 2026 · Sebastian Morgner · info@leadership-munich.org