When Postal Codes Matter More Than Studying: Algorithms as Arbiters of Fair Assessment
When Postal Codes Matter More Than Studying: Algorithms as Arbiters of Fair Assessment

Summary/Abstract: During the Covid-19 pandemic, educational systems around the world became heavily reliant on data-intensive technologies and algorithmic decision-making systems. In the UK, concerns over grade inflation for A-levels qualifications led the overseeing authority to using a grading algorithm in the hopes of providing objective, neutral and fair assessment. In reality, the algorithm displayed bias by downgrading nearly 36% of the grades of students coming from more marginalised groups, resulting in an ‘A-levels fiasco’.We applied a critical discursive psychology approach to explore how students, teachers, parents and policymakers impacted by algorithm-based grading expressed and relied on different senses of fairness (formal, implied contractual, relational, retributive) when commenting on the incident in international news media (N=329).Our findings indicate that while the affected agents evoked and clung to different senses of fairness, the algorithm itself also emerged as a meta-agent of fairness in the discursive constructs of social reality. The analysis further illustrates how public authorities held overly optimistic and techno-solutionist imaginaries about the neutrality and objectivity of AI-based systems, revealing a crucial need for oversight bodies that guide public institutions in responsibly deploying safe and trustworthy data technologies.

• Details
• Contents