AI on the job: translating without erasing occupational differences
In Fribourg’s food sector and bilingual industry, instructions often cross both language and occupational boundaries. The management question is specific: how can we check that an AI rewrite preserves the operational meaning for those who must act on it?
What the literature establishes
Paul R. Carlile (2002, Organization Science) shows, through a study of product development, that occupational knowledge is embedded in different practices and interests. Sharing it therefore involves more than transmitting correctly worded information. Shared artefacts can help occupational groups represent their differences and work through their consequences, rather than make them invisible.
Understanding sometimes requires transforming the explanation
Beth A. Bechky (2003, Organization Science) examines exchanges between occupational communities on a production floor. She shows how understanding develops when an explanation is recast in the context of someone else’s work, notably through concrete objects and problems. The mechanism goes beyond finding a common vocabulary: words must connect with what each occupational group sees and does.
The premature conclusion
One might conclude that AI capable of translating and simplifying resolves these coordination difficulties. These studies do not, however, examine AI and cannot support that conclusion. A fluent rewrite may leave a disagreement about a threshold, an exception or a responsibility untouched; it may also make that disagreement less visible. (our executive and employee training programmes)
What the evidence cannot guarantee
These qualitative studies illuminate mechanisms in specific settings without measuring the effect of automated translation in a bilingual company. They provide neither an expected error rate nor a guarantee that their findings transfer to Fribourg. In practice, checking only a document’s linguistic accuracy cannot establish whether two recipients infer the same action from it.
A practical check in Fribourg
Within « IA on the job », an exercise proposed to a bilingual food or industrial company in Fribourg could use non-sensitive instructions, outside operational use, potentially with support from the local university community. Over two weeks, have each instruction rewritten in French and German, then separately ask the recipient occupational groups to identify the action required, the stopping condition and the person to notify. Measure the proportion of instructions for which all three elements match a reference validated beforehand by the relevant professional leads, and compare this with the result obtained using the existing versions. Keeping the versions and responses makes the check verifiable: the focus of training becomes preserving operational meaning, not merely producing elegant text. To go further: explore the AI on the Job training in Fribourg, or browse our executive and employee training programmes in Switzerland.
In pictures: AI on the Job in Fribourg
- ai on the job
- Fribourg
- research
- training fribourg