Human-Centered AI for Automotive Systems: Towards Explainable, Intercultural, and Standardized Integration

Abstract

The integration of intelligent systems into socio-technical environments requires not only technical excellence but also a systematic consideration of human, cultural, and organizational factors. This paper proposes a Human-Centered Artificial Intelligence (HCAI) framework tailored to the automotive domain, bridging the gap between international standards (e.g., Automotive SPICE, ISO 26262, ISO 9241-221) and the practical deployment of AI-enabled systems. The approach is based on three complementary dimensions: (1) Explainability and Transparency, ensuring that AI-supported decision-making processes are comprehensible to engineers, managers, and auditors; (2) Intercultural Design Integration, incorporating cultural user interface design principles to enhance acceptance in global development teams; and (3) Standardized Assessment, leveraging process models such as Automotive SPICE PAM 4.0 to establish consistent, auditable practices. The research employs context-augmented generation (CAG) techniques with local large language models to assess AI outputs against normative requirements. A multi-agent framework supports evidence extraction, classification, and compliance checking. We introduce Explainable Document Labeling (EDL) to enhance transparency through structured annotations of assessment outputs. Evaluation through industry presentations at the VDA Automotive SYS 2025 Conference, structured demonstrations with automotive suppliers, and systematic reproducibility tests demonstrate that this approach generates standardized outputs with a consistency high enough to work with, addressing one of the major challenges in AI-assisted assessments. Beyond automotive, the findings contribute to understanding how people and intelligent systems can work together effectively in safety-critical industries, illustrating how HCAI principles, intercultural design, and standardized process assessment can jointly advance the reliability, acceptance, and sustainability of intelligent human-systems integration.

Keywords: Human-Centred AI, Intercultural Design, Automotive SPICE, Explainability, Human-Autonomy

DOI: 10.54941/ahfe1006935