Summary
The Centre for Democracy Aarau conducted a pilot study on artificial intelligence and multilingualism in the federal administration on behalf of the Federal Delegate for Multilingualism. The study was presented in Bern on June 16, 2026. It shows that AI tools such as machine translation and large language models can break down language barriers, but simultaneously endanger cultural distinctiveness and weaken understanding between language communities. Imbalances between language groups are exacerbated by varying translation quality—Romansh in particular is underrepresented.
People
- Nicoletta Mariolini (Federal Delegate for Multilingualism)
Topics
- Artificial intelligence and language policy
- Multilingualism in public administration
- Machine translation and LLM
- Swiss Language Ordinance
Clarus Lead
AI penetration of the federal administration is already reality—yet the legal framework lags behind. The study uncovers a strategic dilemma: While AI tools bring efficiency gains to French- and Italian-speaking employees, they simultaneously promote a backward shift toward monolingualism. The danger lies less in the technology itself than in its unregulated use without protective guidelines for Switzerland's linguistic cohesion.
Detailed Summary
Machine translation and large language models have long since become part of daily administrative work and are fundamentally transforming the creation of multilingual content. The technology offers immediate advantages: it lowers entry barriers for French- and Italian-speaking employees, who can now work more efficiently in their native language, thereby increasing attractiveness for members of linguistic minorities in recruitment processes.
However, quality differences between languages are considerable. While German translation capabilities function significantly better, French and Italian fall noticeably behind, and Romansh remains virtually unprocessed. This technical asymmetry structurally reinforces existing imbalances between the four language groups. At the same time, there is a risk that the availability of translation tools leads to less original multilingualism—employees could increasingly communicate monolingually and rely on AI to translate afterward. The long-term effect: language competencies erode, cultural nuances are lost, and language's unifying role for national cohesion weakens.
The Language Ordinance defines target quotas for language representation, yet the regulatory gap in AI use is evident. Mariolini's message is therefore clear: Switzerland must consciously shape AI in favor of linguistic diversity rather than let it happen passively.
Key Messages
- AI tools are already widespread in the federal administration, yet an adapted legal framework is lacking
- Machine translation reinforces existing imbalances between German, French, Italian, and Romansh
- Efficiency gains through AI risk a decline in multilingualism and cultural understanding in the administration
Critical Questions
Data Quality: What sample size and authorities did the pilot study include? Are the results generalizable to the entire federal administration?
Quality Measurement: By what criteria was the translation quality of AI tools assessed between German, French, Italian, and Romansh? Were native speakers included?
Cause and Effect: Does the study provide concrete evidence that AI use leads to less multilingualism, or are these risk analyses based on theoretical grounds?
Regulatory Alternatives: What specific measures did the study concretely recommend besides "completing the legal framework"?
Minority Protection: How can rules for AI use ensure that Romansh and smaller language groups are not further marginalized?
Implementation: Who bears responsibility for implementation—individual authorities or a centralized guideline?
Bibliography
Primary Source: Challenges of Artificial Intelligence for Multilingualism in the Swiss Federal Administration – Plurilingua/Federal Council, June 16, 2026
Supplementary Sources:
- Language Ordinance (SpV) – Federal Chancellery
- Grand Interview: Nicoletta Mariolini – Radios Régionales (French)
Verification Status: ✓ June 16, 2026
This text was created with the support of an AI model.
Editorial responsibility: clarus.news | Fact-check: June 16, 2026