Summary

The National Research Program NFP 77 examined the effects of digitalization on Swiss society, labor market, and democracy in 46 projects (2020–2025). Program director Abraham Bernstein warns: Digital transformation is not purely a technological phenomenon, but an interplay between technology and social actors. The research identifies Swiss strengths in AI research but reveals deficits in media literacy, gender equality, and regulation. On Wednesday, the research consortium presented 12 key recommendations for politics, business, and administration.

Persons

  • Abraham Bernstein (Computer Science Professor University of Zurich, Director of Steering Committee NFP 77)
  • Mark Eisenäcker (Researcher University of Zurich, Democracy Studies)

Topics

  • Digital Transformation
  • National Research Programs
  • Media Literacy and News Consumption
  • Labor Market Changes
  • Artificial Intelligence and Regulation
  • Gender Parity in Tech Competencies

Clarus Lead

Switzerland stands at a critical turning point: While research and development capacities are internationally competitive, information loss and digital divide threaten to endanger the functionality of direct democracy. With 48% "news-deprived" citizens—especially among young people—a cornerstone of the Swiss voting system is losing substance. In parallel, massive gender gaps are evident: Women rate their digital competencies significantly lower than men and participate less actively in technological transformation. The research therefore demands a paradigm shift: not regulatory competition with the EU, but decentralized control by individuals, employers, and government institutions.

Detailed Summary

Digitalization is often treated as a technological phenomenon. However, Bernstein argues for a relational understanding: technology and society influence each other mutually. Concrete examples illustrate the deep-reaching effects—from the decline of printed media (commuters using smartphones instead of newspapers) to streaming services to the reshaping of office work and classroom instruction. These structural changes require not only technical but above all design knowledge.

In education, NFP 77 studies recommend lifelong competency training beyond formal schooling. Employers and employees share responsibility for continuous further education—an approach suited to rapid development (Bernstein updates his AI lecture slides weekly). On the labor market, two pilot projects show solutions: Instead of job titles, competency platforms could better match job seekers with actual job requirements—for example, allowing a clockmaker graduate to apply their skills in technical niche areas hidden under traditional job titles.

Democracy research identifies an existential risk: incomplete news consumption (48% news-deprived) undermines informed voting. Two models demonstrate countermeasures: A municipality in Canton Bern conducted electronically supported municipal discussions—thousands discussed before votes, which increased sense of participation and legitimacy. A participatory budget (University of Freiburg / ETH) in a city enabled direct shaping of investments. Such digital tools work only if they are transparent and fair—a critical question with artificial intelligence.

AI systems (language models, decision trees) are statistical, not logically deterministic. They produce errors and "hallucinations" (fabricated facts). Bernstein does not advocate for trust, but for reliability knowledge: users must learn to recognize contexts in which AI outputs are reliable—more often with general knowledge, rarely with medical dosages. This requires critical skepticism and continuous boundary exploration.

Switzerland pursues a decentralized regulatory approach rather than the EU model (AI Act). Bernstein recommends: define clear red lines (what AI may and may not do), but also create enforcement mechanisms—otherwise rules are ignored, such as with deepfakes in election campaigns. Central is the role of universities: In collaboration with partners, Switzerland is developing its own language model to maintain technological sovereignty.

Gender and age differences reveal who benefits from digitalization: women rate their competencies lower and invest less time in digital learning; older people are absent from higher education adaptations (universities use AI, but older faculty hesitate). A playful, error-tolerant approach to technology would be particularly beneficial here—because statistical systems inevitably make mistakes.

Key Messages

  • Digital transformation is an interplay between technology and society; passive participation is not politically sustainable.
  • 48% of the population no longer consume news (news deprivation), jeopardizing informed voting.
  • Competency platforms can improve job matching by aligning skill profiles rather than job titles.
  • Digital participation tools (e-voting, participatory budgeting) demonstrably increase voting legitimacy—when transparency is assured.
  • Artificial intelligence requires reliability knowledge, not trust; users must know error margins.
  • Women and older people participate less and underestimate their competencies—structural inequality must be actively combated.
  • Switzerland is developing its own language model for technological sovereignty; regulation should set red lines without blocking innovation.

Critical Questions

  1. Evidence/Data Quality: The study cites "48% news-deprived" among young people—is this figure based on the Eisenäcker study itself, and what is the precise definition (zero news consumption vs. reduced usage)?

  2. Conflicts of Interest: Bernstein directs the research program and presents its findings himself. How were independent external reviews conducted to exclude confirmation bias?

  3. Causality: The statement "women rate their competencies lower and participate less"—is lower self-assessment a cause or consequence of market exclusion / underrepresentation in tech teams?

  4. Media Literacy Feasibility: If the research recommends "systematically inspiring youth about politics" rather than media directly—how concretely does this motivation strategy work in heterogeneous classrooms?

  5. AI Regulation Switzerland: The report criticizes insufficient enforcement of rules (deepfakes in election campaigns). What resources and administrative structures would need to be built to monitor this?

  6. Labor Market Matching: The job platform model ("competencies instead of titles") assumes employers standardize job profiles. How realistic is this in a federal system with different industry definitions?

  7. Digital Participation Scaling: The pilot studies (e-voting in one municipality, participatory budgeting in one city) show positive effects—but how do these scale to national votes without information overload?

  8. Language Model Sovereignty: Switzerland is developing its own language model—with what timeline, budget, and data protection standard will this be realized, and will it remain competitive against OpenAI/Google/Meta?

Additional Reports

No additional reports available from source.

Source Directory

Primary Source: Daily Conversation with Simone Hulliger (SRF), 28.05.2026 – National Research Program NFP 77 "Digital Transformation": https://download-media.srf.ch/world/audio/Tagesgespraech_radio/2026/05/Tagesgespraech_radio_AUDI20260528_NR_0016_b6221e78ce254b96a8911ded69daf9cd.mp3

Supplementary Sources / Cited Studies:

  1. Stanford AI Index (2026) – AI Research Capacities per Capita
  2. Eisenäcker et al. (University of Zurich) – News Deprivation Among Young People
  3. University of Bern – E-Voting and Participation (NFP 77 Project)
  4. University of Freiburg / ETH – Participatory Budgeting (NFP 77 Project)
  5. University of Basel – Gender Differences in Tech Self-Assessment (NFP 77 Project)

Verification Status: ✓ 28.05.2026

This text was created with the support of an AI model.
Editorial Responsibility: clarus.news | Fact-Checking: 28.05.2026