Examining the automated systems that affect real lives
Algorithmic decision-making now affects access to employment, credit, insurance, healthcare and government services. The Algorithmic Audit runs technical evaluations of these systems, seeks right-of-reply from vendors and publishes full methodology alongside every finding. We are not interested in claims. We are interested in outcomes.
We submitted 1,200 synthetic loan applications to four major credit assessment tools, varying only postcode and income source. The outcome differentials were statistically significant and correlated with socioeconomic proxies.
Three enterprise AI hiring tools claimed independent bias audits were technically infeasible. We ran the audit using documented methodology. The systems were not bias-free.
Automated systems assist with benefits eligibility decisions in three Australian states. We filed FOI requests, engaged statistical experts and documented what accountability mechanisms exist. The answer was: few.
Direct discrimination in insurance pricing is prohibited. Pricing models that use variables which function as proxies for protected characteristics are a legal grey area. We mapped the proxies.
+5 more articles in this series. Subscribe to the newsletter to receive every new instalment.
Every new article in The Algorithmic Audit goes straight into the weekly newsletter. No separate subscription required.
No spam. No affiliate links. Just honest analysis. Unsubscribe any time.