evidoria

Browse practices

Topic: AI in public services clear · back to the knowledge map

This selection at a glance

209
matching practices
3%
strong evidence, of 209 assessed

Evidence mix

Randomised controlled trial 1
Quasi-experimental 6
Observational / pre–post 79
Descriptive / self-reported 123

Most practices in: United Kingdom (9) · United States of America (8) · Singapore (7)

Results

209 result(s) Download CSV Get alerts for this search Cards List
Good practice Evidence: Observational / pre–post Top 98% 20/100

Försäkringskassan's discriminatory welfare-fraud AI (Sweden)

Sweden's Social Insurance Agency used an ML risk-scorer from ~2013 to detect benefit fraud; a 2024 investigation found it disproportionately flagged women, …

Stockholm, Sweden