If an Algorithm Feels Safer Than the Leader, Listen

An employee notices that a proposed target rests on incomplete information. In front of a human manager, the employee weighs more than the quality of the evidence. Will the question be taken personally? Will disagreement be remembered at appraisal time? Does the manager already favour another colleague? Is respectful silence safer than professional candour?

Place the same concern before an apparently impartial system and some of that interpersonal calculation may disappear. A machine has no wounded pride, no friendship to protect and no visible status to defend. It may therefore feel easier to challenge—even when the system itself is opaque, biased or unable to understand the situation.

That possibility creates a serious leadership test. If people feel safer raising a work-related concern to an algorithm than to the person entrusted with authority, the solution is not to automate every relationship. It is to examine what human leadership has made dangerous.

Human leaders earn trust when they make fairness visible, receive inconvenient knowledge without retaliation and ensure that both human and automated decisions can be questioned, explained and corrected.

A Scientific Reports article published on 21 July 2026 reports three scenario-based experiments involving participants recruited in China. Across the experiments, participants expressed greater willingness to speak up to an algorithmic leader than to a human leader. The difference appeared in cognitive tasks but not emotional tasks.