AI on the Job training in Geneva · canton of GenèveDoes generative AI blur the message when your team speaks two languages?

Multicultural teams constantly navigate between two or three languages, where the shift from French to English carries managerial meaning. When generative AI drafts a message for such a team, it often erases the markers that signal to the recipient how to interpret it.

What research establishes about multilingual message coherence

Albert Mehrabian and Susan Ferris (1967, Journal of Consulting Psychology) demonstrated that message meaning depends heavily on three simultaneous channels: verbal content, tone, and context. In a multilingual team, the choice of language itself is a contextual signal that orients interpretation. When a manager writes an organizational directive in French but commercial urgency in English, they create a perceptible distinction. Generative AI, trained to produce formal clarity, tends to neutralize these subtle differences in favor of stylistic uniformity that can mask real intentions.

The mechanism: AI produces formal coherence, not situational coherence

Kees de Dreu (2002, European Journal of Work and Organizational Psychology) demonstrates that high-performing teams process information contextually — they integrate frame and indirect signals, not just explicit content. A generative AI, trained on statistical regularity, produces uniform formulation: it 'normalizes' French and English toward neutral register, erasing the deliberate gaps a manager used to nuance their message across languages or stakes. The message becomes more formally readable, but loses the layers of interpretation that the multilingual team naturally activated.

The hasty conclusion and its correction

One might conclude: AI makes communications less effective in multilingual context. But observation requires correction: AI improves transactional clarity (factual content, instructions, figures) precisely because it removes ambiguities. Conversely, it impoverishes messages where intention rests on shared code, tone, or calculated language switching. The loss is never total; it is selective. Formal exchanges gain; relational communications lose. (our executive and employee training programmes)

Limitations: laboratory context versus field reality

Most studies on multilingualism in organizational context are conducted with expatriates or research teams, rarely with financial services or trading professionals where language choice is a daily strategic act. Moreover, data on AI and multilingualism remain limited; we lack studies comparing the effect of an AI-generated message in French versus English versus hybrid on a team's comprehension and engagement. Tool adoption is too recent to have robust empirical measures of this specific phenomenon.

A practical check in Genève

A manager of a multilingual team can test the hypothesis this way: draft an important message without AI, then ask the tool to make it 'clearer' and compare both versions. Measure concretely: how many English passages did the AI version preserve intact? How many language transitions did it 'correct'? Then send each version to a colleague from the same teams (private bank, international organization, luxury watchmaker) and ask: where do you see the urgency level? Where is the shift? These comparisons quickly reveal whether AI erases the signals only the bilingual team grasps. Tracking three critical messages shows whether the gain in readability offsets the loss of intention. AI training in the canton of Geneva. To go further: explore the AI on the Job training in Geneva in the canton of Geneva, or browse our executive and employee training programmes in Switzerland.

In pictures: AI on the Job in Geneva

• ai on the job
• Geneva
• research
• training geneva