Training managers to manage in an AI-augmented workplace: where to start?

What does AI change in a manager's work?

Uneven productivity gains. A study published in the Quarterly Journal of Economics (Brynjolfsson, Li and Raymond, 2025) followed 5,172 customer support agents equipped with a generative AI assistant. The number of issues resolved per hour rose by 15% on average, with wide differences: the least experienced agents gained in both speed and quality, while the most experienced gained little speed and lost slightly in quality. Two people in the same job can therefore get very different benefits from the same tools.

Freed-up time that does not reallocate itself. When a task is accelerated, the time saved is not automatically reinvested in higher-value tasks. It is up to the manager to decide with the team how to use it.

Questions of responsibility. When an employee relies on a recommendation produced by AI, the manager needs to set clear rules: which outputs must be checked, by whom, and who validates the final decision. For high-risk systems, the AI Act requires human oversight to be assigned to people with the necessary competence, training and authority (Article 26).

Fairness issues. If an employee produces faster thanks to AI, the manager has to choose: give them more work, train the others to close the gap, or rethink how roles are distributed.

Which management skills to develop?

Assessing the quality of AI outputs. A manager who cannot spot an error in a document produced with AI is in practice delegating their oversight to the tool. This skill corresponds to checking information, at the heart of the AI literacy expected under Article 4 of the AI Act.

Redefining objectives and performance measurement. When AI takes over part of the work, volumes produced become a less reliable indicator. The quality of decisions, the relevance of checks and the value added to the raw output carry more weight.

Addressing concerns. Some employees see AI as a lever, others as a threat to their job or expertise. Managers can only respond usefully if they themselves understand what the tools do and do not do in their team's jobs.

Maintaining the human connection. Building trust, spotting signs of disengagement, giving meaning to decisions: these skills are not new, but they take up more room when part of the work is automated.

What legal framework should managers know?

In French companies with at least 50 employees, the introduction of new technologies is one of the subjects on which the CSE (the employee representative body) is informed and consulted (Article L2312-8 of the Labour Code). For a high-risk AI system used in the workplace, the employer must also inform workers' representatives and the workers concerned before putting it into service (Article 26(7) of the AI Act). These obligations lie with senior management, but managers should know they exist in order to answer their team's questions.

Which training formats suit managers?

Managers have little time and expect answers to concrete situations. Three formats lend themselves to this:

• Short sessions based on real cases: each manager brings a situation they have experienced, analysed as a group to draw principles for action.
• Peer co-development: small groups of managers meet regularly to share practices and build responses suited to the organisation's context.
• Reverse mentoring: an employee comfortable with the tools helps their manager get to grips with them.

What does not work?

Training that explains how "AI will transform work" without starting from managers' own situations mostly produces anxiety. Managers need precise answers: how to assess an employee who uses AI heavily, how to react to a refusal to use the tools deployed, how to set objectives that take productivity gains into account without penalising those less at ease with these tools.