IA on the job: does rephrasing a disagreement resolve it?
Asking AI to rephrase a tense message is a plausible use of assistance at work. The management question is specific: how can we check that rephrasing helps address the disagreement rather than making it less visible?
What the literature establishes
Karen A. Jehn (1995, Administrative Science Quarterly) distinguishes relationship conflict from conflict over the substance of work and shows that their effects vary, notably with the nature of the tasks. Her study does not support treating every disagreement as a dysfunction: discussing alternatives can be useful under certain conditions. For a manager, reducing tension and addressing a difference of opinion are therefore distinct objectives.
Technical disagreement does not always stay technical
Carsten K. W. De Dreu and Laurie R. Weingart (2003, Journal of Applied Psychology) report, in their meta-analysis, negative associations between both forms of conflict and performance as well as member satisfaction. The negative association between task conflict and performance is weaker when task conflict is less closely linked to relationship conflict. These findings suggest examining how disagreement over work combines with interpersonal tensions, rather than assuming that technical debate is necessarily beneficial.
The premature conclusion
One might infer that AI which removes accusatory wording necessarily improves cooperation. Yet these studies test neither AI-generated reformulations nor their effects on conflict resolution. A message can become more courteous while leaving an incompatibility between the requested deadline, required checks and available resources untouched. (our executive and employee training programmes)
What current practice does not measure
Assessing a reformulation solely by its tone measures the apparent quality of the exchange, not how the problem is addressed. The cited studies concern conflict within groups and provide no causal evidence about the effectiveness of a writing assistant. A meaningful assessment must therefore distinguish perceived easing of tension, understanding of the positions and the existence of a decision that is actually accepted.
A practical check in Neuchâtel
In an « IA on the job » workshop in Neuchâtel, a role-play could start with a disagreement over batch release, adapted to microtechnology, watchmaking or precision microelectronics. Before any rephrasing, each party separately records their non-negotiable constraint and the decision they want made; after reading the revised message, the other party must restate both without consulting that record. The measure is the proportion of constraints and requested decisions correctly restated, checked against the original records and compared with a similar exercise without AI. This test does not demonstrate that a conflict has been resolved, but it can check whether greater courtesy preserves the information needed to address it. To go further: explore the AI on the Job training in Neuchâtel in the canton of Neuchâtel, or browse our executive and employee training programmes in Switzerland.
In pictures: AI on the Job in Neuchâtel
- ai on the job
- Neuchâtel
- research
- training neuchâtel