On April 1st, I published an April Fool's joke. A fictional company hands over full leadership responsibility to its artificial intelligence (AI) – and the AI voluntarily resigns after seven days. Some people have written to me: „Exciting! Where can I learn more about this experiment?“  Other reactions were simply: „Cool topic!“.

The experiment is fictional. But the dynamics I described in the article are not. When I wrote the article, it seemed absurd to me. Yet many people believed the experiment was real. This surprised me at first, and then made me thoughtful, but then I wasn't surprised at all. What was described is now so close to lived reality that it is no longer perceived as absurdity. That is the real joke. And it is more serious than the prank itself.

The starting premise is real

The idea of automating leadership is no longer just a thought experiment. It's embedded in every discussion about data-driven decision-making, algorithmic management, and AI-assisted process optimization. And it's based on an assumption that is rarely openly stated: that humans are the unreliable factor in the system.

When decisions are made too slowly, when changes don't take hold, when teams don't get on board, the implicit explanation often lies in the fact that anybody the wrong attitude, too little consistency, or not enough speed. The solution follows this logic: better control, clearer guidelines, less human discretion.

An AI then appears as a consistent further development of this idea.

2. The contradictions within the system are real

In the April Fool's joke, the AI analyzes the company's requirements and discovers that they are mutually exclusive:

• Be efficient – and take time for quality.
• Take responsibility – and stick to the guidelines.
• Be innovative – and don't make mistakes.
• Think along – but don't decide on your own.

This is not satire. This is a very accurate description of tensions that are valid in many organizations simultaneously. They cannot be resolved by ignoring them or optimizing faster. They belong to the structure of social systems.

3. The Limits of AI are Real

The crucial moment in the text is not the resignation. It's the sentence before it:

„I am not part of this system."

An AI can analyze, structure, recognize patterns, and derive decisions. What it cannot do: experience the dynamics of a social system from within. It cannot feel when trust erodes. It cannot perceive when withdrawal arises behind formal agreement. It cannot become part of the tension it's trying to manage.

This is not a weakness of a specific AI. It is a structural property.

Leadership in social systems doesn't just work through analysis and optimization. It requires participation, being involved in the same reality that everyone else is also working in.

4. What follows

Organizations are not machines. They do not function according to a single logic that one only needs to set correctly. They are contradictory, dynamic, and not entirely predictable, and this is not their deficit, but their mode of operation.

Good leadership in this context does not mean resolving all contradictions. It means remaining capable of action, even when you can't resolve them. Perceiving what is happening in the system. Setting impulses and observing what emerges from them. And understanding one's own role not as a controlling authority, but as part of the system.

The AI in my April Fool's joke recognized that and drew the conclusion. The question that remains is: Will we?