What happened
Arbeidstilsynet, the Norwegian Labour Inspection Authority, began developing Prediksjonsindekstjenesten in 2017. Its current version has been in use since 2021. The model was trained on the results of earlier inspections, and it processes no personal data. It estimates the probability that an inspection of a business will find serious violations. Inspectors see this as one of four categories, from low to high, beside other information about the business. Riksrevisjonen, the Office of the Auditor General, studied the system as one of four AI cases in a report dated 2 September 2024. It reports that the authority can point to more accurately targeted and thus more effective inspections. It writes that inspecting businesses in higher risk groups raises the chance of finding more and more serious violations. The auditor relays the authority's own assessment of these results. It notes that the system has not been systematically evaluated. The auditor gives no hit rates or counts of violations found.
Sources
-
Report, by authors with no tie to the system.
Arbeidstilsynets bruk av KI som beslutningsstøtte ved utvalg av virksomheter for tilsyn kan vise til mer treffsikre og derav effektive tilsyn. For eksempel ved at valg av tilsyn med virksomheter i høyere risikogrupper øker sannsynligheten for å avdekke flere og mer alvorlige lovbrudd.
Details
AI system
- Workflow
-
The model was trained on the results of earlier inspections. Inspectors saw a risk category for each business beside other information about it.
Follow-up
- Conditions for replication
-
Inspectors decide which businesses to inspect. The index is not used for every kind of business, such as farms without employees.
Entry details
- Sector
-
Labor inspection.
- Purpose
-
Arbeidstilsynet built Prediksjonsindekstjenesten to estimate the probability of serious violations at a business so that inspectors can choose which businesses to inspect.
Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Prediksjonsindekstjenesten helps Norway aim its labor inspections (AIBD-0710). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0710.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)