Why Workers Use AI Less Often: Reasons for Declining Workplace AI Use in Germany
Citation. Lünich, M. (2026). Why Workers Use AI Less Often: Reasons for Declining Workplace AI Use in Germany (Factsheet No. 18, Opinion Monitor Artificial Intelligence). SocArXiv. https://doi.org/10.31235/osf.io/pjn8x_v1
Summary. This public preprint reports a 2026 survey of German workers about recent changes in workplace AI use. Most respondents said their use had not changed, more reported an increase than a decrease, and people who reduced use most often pointed to weak practical value, unreliable output, data-protection uncertainty, and unclear workplace rules. It provides a descriptive snapshot of friction after AI is available, not evidence that any one factor causes people to stop using it.
Key claims
- Among 2,585 respondents with a valid answer, 65.6% reported no recent change in workplace AI use, 28.6% reported increased use, and 5.8% reported decreased use over the prior three months.
- Among workers who had reduced their use, the most strongly endorsed reason was that AI brought no noticeable added value (63.4% agreement); uncertainty about data-protection requirements and unreliable results were also common reasons.
- Reduced use was associated with less positive views of AI, lower interest, lower self-rated AI competence, lower job satisfaction, and lower identification with the organization, but the report cannot establish which came first.
- Workers in higher-skill occupational groups were more likely to report increased use, while workers in lower-skill groups were comparatively more likely to report decreased use.
Evidence & limitations
- Quota-based online survey of the German working online population, fielded 29 April–11 May 2026; 3,012 people completed the survey and 2,585 were included in the change-in-use analysis.
- The report is descriptive and asks respondents to recall change over three months, so it does not establish causal effects or exact transitions in AI use.
- The reported reasons for reducing use come from 149 respondents, and measures of use, attitudes, competence, and work experience are self-reported.
- Public SocArXiv provenance was verified through the OSF API: its primary-file SHA-256 exactly matches the ingested Drive PDF. This is a public preprint, not peer-reviewed evidence.