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

Jul 16, 2026

Upload by Ophelia

Explainability has emerged as a critical requirement for AI-based systems, especially in safety-critical and regulated settings. Prior work proposes frameworks and methods, yet limited empirical evidence shows how existing Requirements Engineering (RE) practices support explainability requirements across the full RE lifecycle, particularly in industry. This paper presents early results from a qualitative, multi-phase industry study with eight Daimler Truck practitioners using think-aloud protocols and group discussions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages5

Reading Time13 min

Abstract

Keywords

Introduction

Evaluating RE Practices for Explainability: Synthesizing Insights from Daimler Truck into an Explainable RE Framework Proposal Umm-e-Habiba [umme.habiba@tum.de](umme.habiba@tum.de)[ ](umme.habiba@tum.de)TUM School of CIT, Technical University of Munich Heilbronn, Germany Lucas Mauser Daimler Truck AG, Leinfelden-Echterdingen Germany [lucas.mauser@daimlertruck.com](lucas.mauser@daimlertruck.com) Jonas Fritzsch Institute of Software Engineering, University of Stuttgart Stuttgart, Germany [jonas.fritzsch@iste.uni-stuttgart.de](jonas.fritzsch@iste.uni-stuttgart.de) arXiv :2607 . 1 177 1v 1 [ cs . SE] 13 Jul 2026 Justus Bogner Department of Computer Science, Vrije Universiteit Amsterdam Amsterdam, Netherlands [j.bogner@vu.nl](j.bogner@vu.nl) Abstract Explainability has emerged as a critical requirement for AI-based systems, particularly in safety-critical and regulated domains. Although prior research has proposed frameworks, patterns, and usercentered approaches to support explainability, there is limited empirical understanding of how existing Requirements Engineering (RE) practices support explainability requirements across the RE lifecycle, especially in an industrial context. This paper reports early ﬁndings from an ongoing industry-based study investigating how explainability requirements are elicited, speciﬁed, and validated using established RE techniques. We conducted a multi-phase qualitative study with eight practitioners at Daimler Truck, employing think-aloud protocols and moderated group discussions across requirements elicitation, speciﬁcation, and validation steps. Our preliminary analysis reveals recurring challenges across all steps, including conceptual ambiguity during elicitation, limited testability and expressiveness during speciﬁcation, and fragmented validation due to vague criteria and regulatory uncertainty. These ﬁndings indicate that current RE practices provide limited support to systematically address explainability requirements. The paper contributes empirical insights into step-speciﬁc and cross-cutting challenges and outlines a research vision toward developing an empirically grounded RE framework for explainable AI-based systems. Keywords Explainable AI, Requirements Engineering, Framework Proposal, User Study, Focus Group 1 Introduction Artiﬁcial intelligence (AI) is increasingly used in domains such as medicine [6, 18], law [17], autonomous driving [3], and loan application approval [24] . In these and other high-stakes settings, AIbased systems often support or inﬂuence critical decision-making processes. Consequently, responsible users, including physicians, judges, drivers, and bankers, require varying levels of explanation to appropriately trust, assess, and act on AI outputs. Ensuring that AI decisions are interpretable is essential for safe, responsible, and legally compliant AI deployment [21, 22] . Stefan Wagner TUM School of CIT, Technical University of Munich Heilbronn, Germany [stefan.wagner@tum.de](stefan.wagner@tum.de) From a software engineering perspective, developing explainable AI-based systems requires eﬀective methodologies to elicit, specify, and validate explainability-related requirements. However, as explainability has only recently emerged asa key non-functional requirement [19], practitioners currently receive limited support for systematically integrating explainability into Requirements Engineering (RE)processes. Explainability poses particular challenges for RE due to its context-dependent, multi-dimensional, and humancentered nature. Prior research has increasingly investigated explainability from an RE perspective by proposing conceptual frameworks, experi ence reports, and user-centered methods that supportspeciﬁcexplainability related activities [5, 29, 30] . In parallel, several studies have explored explainability requirements in real-world contexts through interactive and contextual user studies, focusing on parti

FAQs

What problem does the paper address about explainability and requirements engineering?

It investigates how existing Requirements Engineering practices help engineer explainability requirements across elicitation, specification, and validation, especially in an industrial setting. Prior research is described as lacking holistic, end-to-end empirical understanding.

How was the study conducted at Daimler Truck?

The study used a multi-phase qualitative approach with eight practitioners at Daimler Truck. Practitioners followed think-aloud protocols and participated in moderated group discussions across the RE steps.

What challenges were found in elicitation, specification, and validation?

Recurring challenges include conceptual ambiguity during elicitation, limited testability and expressiveness during specification, and fragmented validation due to vague criteria and regulatory uncertainty. Overall, current RE practices provide limited systematic support for explainability requirements.

Jul 16, 2026

Upload by Ophelia