Reliability and security of AI models in hardware systems: The role of expert knowledge and technical talent management

The integration of artificial intelligence into modern hardware systems introduces new challenges related to reliability and security. Unlike traditional deterministic designs, AI-based hardware systems exhibit adaptive and partially non-deterministic behaviour, which complicates verification and risk assessment processes. While AI-assisted tools significantly improve efficiency in hardware verification, their effectiveness remains limited in complex or boundary operating conditions. This paper examines the role of human expertise and technical talent management in ensuring the reliability and security of AI-based hardware systems. The study combines insights from existing literature with an industry-based case example from digital hardware verification practice. Particular attention is given to the limitations of AI-assisted verification tools and the importance of expert judgment in interpreting results and preventing incorrect design decisions. The paper further argues that technical talent management, including continuous learning, mentorship and knowledge retention should be treated as a reliability mechanism rather than a purely organizational or human resource's function. The findings suggest that the quality and trustworthiness of AI-based hardware systems depend not only on algorithms and tools but also on how technical teams are developed, guided, and supported within organizations.