Is your AI Model Accurate Enough? The Difficult Choices Behind Rigorous AI Development and the EU AI Act

A research paper challenges the common view of AI accuracy as purely technical, arguing it involves context-dependent normative decisions that determine error priorities and risk distribution. The study analyzes the EU AI Act's "appropriate accuracy" requirements and identifies four critical choices in performance evaluation that embed assumptions about acceptable trade-offs.

• →AI accuracy evaluation is fundamentally context-dependent and involves normative decisions rather than being purely technical.
• →The EU AI Act mandates "appropriate accuracy" for high-risk AI systems but leaves implementation open to interpretation.
• →Four critical choices shape accuracy assessment: metric selection, metric balancing, representative data measurement, and acceptance thresholds.
• →Technical implementation of accuracy requirements embeds implicit assumptions about acceptable risks and errors.
• →The research provides practical guidance for regulators, auditors, and developers implementing AI safety requirements.