Evaluating RE Practices for Explainability: Synthesizing Insights from Daimler Truck into an Explainable RE Framework Proposal

Explainability requirements are poorly understood in practice: The study of eight practitioners at Daimler Truck revealed "conceptual ambiguity during elicitation," meaning teams struggle to agree on what "explainability" actually means when building AI systems for trucks—a safety-critical domain where clarity is essential.

Current engineering practices aren't equipped for AI explainability: Established Requirements Engineering techniques fail to adequately support explainability across the full development lifecycle, with particular gaps in "limited testability and expressiveness during specification," making it difficult to write clear, measurable requirements for how AI systems should explain their decisions.

Validation of explainability remains fragmented and uncertain: Organizations face challenges validating whether AI systems actually meet explainability requirements due to "vague criteria and regulatory uncertainty," meaning companies like Daimler Truck cannot confidently confirm their AI systems are sufficiently explainable even after development is complete.

The problem spans the entire development process: Rather than being isolated to one stage, explainability challenges appear "across all steps" of requirements elicitation, specification, and validation, suggesting the issue requires systematic solutions throughout the engineering lifecycle, not just quick fixes at one stage.

Industry needs tailored frameworks for explainable AI: The paper calls for developing "an empirically grounded RE framework for explainable AI-based systems," indicating that generic AI governance approaches are insufficient and that safety-critical industries need specialized methods designed specifically for managing explainability requirements.