Automated Decision-Making and Equality Law

Covering Assessment Criterion 2.2 of FREE08, this video addresses two interlocking areas of law that govern how AI may be used in HR decision-making: UK GDPR and the Equality Act 2010. On the data protection side, the video focuses specifically on Article 22 of UK GDPR and the rights it confers on individuals when automated processing produces decisions with legal or similarly significant effects. Recruitment, promotion, performance ratings and dismissal are all examined as contexts where those rights become directly relevant, alongside the conditions under which automated decision-making may lawfully proceed and what meaningful human intervention must look like in practice.

The equality law dimension centres on indirect discrimination under the Equality Act 2010 and the mechanisms through which AI systems can produce discriminatory outcomes without any discriminatory intent. The video identifies and names several distinct forms of algorithmic bias — including historical pattern perpetuation, proxy discrimination and feature selection bias — and discusses how each arises from the relationship between AI systems and the data on which they are trained. The example of continuous employment history as a variable that may disadvantage women who have taken career breaks for caring responsibilities is among the concrete illustrations used to show how bias enters AI outputs through apparently neutral design choices.

The video also addresses organisational responsibility, making clear that liability for discriminatory AI outcomes does not transfer to third-party vendors and examining what due diligence, equality impact assessments and ongoing output monitoring look like in this context. Human oversight is treated not as an optional safeguard but as a legal requirement, and the distinction between token review and genuinely authoritative human intervention runs throughout the discussion of both UK GDPR compliance and equality law obligations.