“The Trust in AI-Mix: Technical, Legal and Ethical Requirements as well as Psychology”
Jan Spilski, Anja Spilski, Thomas Lachmann · 2025
The increasing spread of artificial intelligence (AI) applications underscores the necessity for an interdisciplinary initiative to ensure the trustworthiness of AI and its alignment to human values and security. A distinction is made between trustworthy AI and trust in AI. The former is defined by adherence to ethical, technical, and legal standards and describes an implementation perspective. The latter reflects a psychological perspective and focuses on the perception of users. For the implementation of trustworthy AI, ethical guidelines from institutions and companies, technical challenges in terms of explainability, robustness, and bias avoidance, and legal requirements are discussed. The user perspective is considered in two respects; first drawing from implementation guidelines provided by the human-centered AI approach. Second, while focusing on the perceptual perspective of trust in AI, findings of developmental psychology, cognitive science, and human–computer interaction are discussed. By connecting technical implementation, ethical and legal requirements, as well as psychological insights, we propose a comprehensive and holistic "trust-in-AI mix" to guide the development of AI systems that foster user trust in AI.