Weshalb die meisten Betriebe Künstliche Intelligenz (noch) nicht nutzen / Why most firms do not use AI (yet)
B · Useful, limited IAB researcher Christian Kagerl reports representative IAB Establishment Panel data on why firms skip AI; credible and useful, but a short post with self-reported reasons and no sample or model details shown.
Summary based on the full text·Original: about 7 minutes to read
In short
An IAB-Forum article by Christian Kagerl of Germany's Institute for Employment Research. One in four German establishments used generative AI in the second half of 2025, but about two thirds neither used nor planned to use any AI. Almost 90% of these non-users say they see no use case. That share barely varies with how exposed their workforce is to AI, which the author reads as unrecognised potential rather than informed rejection.
Key numbers
- 1 in 4German establishments using generative AI in H2 2025, about five times 2023
- ~2/3establishments neither using nor planning to use AI in H2 2025
- almost 90%non-users citing no suitable use case
- 7%non-users who tried AI and were disappointed
- 13% vs 41%AI use rate in establishments with the lowest vs highest occupational AI exposure
As reported by the source.
English·Web page·opens iab-forum.de
What it reports
- Generative AI use: one in four establishments in H2 2025, about five times 2023; a further 9% planned to use AI.
- About two thirds used no AI and had no plans.
- Reasons given by non-users: no use case (almost 90%), lack of knowledge (36%), lack of skilled staff to implement (34%), incompatible hardware or software (33%), data protection and security (32%), legal uncertainty (32%), too expensive (30%).
- Only 7% had tried AI and been disappointed; in information and communication the share is nearly a quarter.
- Lack of use cases is cited by fewer than two thirds of establishments with 200+ staff but almost 90% of micro firms.
- 91% of the least exposed and 85% of the most exposed non-users see no use case, while use rates across all establishments run from 13% to 41% by exposure.
- Internal technical and organisational barriers are cited up to 15 points more often in manufacturing than in services, and 5-10 points less often by younger establishments.
- Associations hold in multivariate regressions controlling for age, industry, works council and staff structure.
How they got it
IAB Establishment Panel survey for H2 2025, with occupational exposure based on each establishment's mid-2022 occupation mix and the Eloundou et al. (2024) measure. Headline adoption figures come from IAB-Kurzbericht 08/2026 by Friedrich and Kagerl.
Limits
- Reasons are self-reported and respondents could tick several (three on average).
- Sample size, weighting and regression output are not shown in the post.
- The survey predates wider agentic tools; the author expects reasons to shift.
- The page shows two different citation dates (20 October and 15 April 2026).
Why it matters for companies
The barrier for most non-adopting firms is identifying a task to apply AI to, more than cost or regulation. Use cases are often firm-specific and found by trying, as the article's example of scaffold-erection firms shows (quotes, billing and scheduling, not the physical work). For suppliers and internal AI teams, concrete task-level examples matter more than general capability pitches. Older firms and manufacturers face more integration friction and should budget for it.
Key passages in translation
Quoted in English
Mit Abstand am häufigsten gaben die befragten Betriebe an, derzeit keine Anwendungsmöglichkeit für KI zu sehen (fast 90 Prozent der Nichtnutzer).
By far the most common answer from the surveyed establishments was that they currently see no use for AI (almost 90 percent of non-users).
Quoted in English
Lediglich 7 Prozent berichten, KI ausprobiert zu haben, ohne dass die Erwartungen erfüllt wurden.
Only 7 percent report having tried AI without their expectations being met.
Quoted in English
Daher empfiehlt es sich für die Betriebe, KI zumindest experimentell zu erproben und so deren betriebswirtschaftliches Potenzial auszuloten.
Establishments are therefore advised to at least trial AI experimentally and so gauge its business potential.
Summary based on the full text. Read the original for the full argument and data.Read the original on iab-forum.de
- Tags
- AI adoption