Building Trustworthy AI
Lars Meloe, Principal Consultant, Digital Trust, DNV
In this thoughtful and deeply practical session, Lars Meloe, Principal Consultant in Digital Trust at DNV, explored the critical requirements for building AI systems that are not only intelligent but also safe, explainable, and aligned with human operators. As AI becomes increasingly embedded in industrial and energy operations, Meloe emphasised that trust must be earned—not assumed.
The presentation addressed a pressing reality: while AI is evolving quickly and expanding in use across the energy sector, it also introduces significant complexity in development, qualification, and operation. Meloe highlighted the growing challenges in human-AI interaction, noting that as systems become more capable, they also become harder for people to understand, monitor, or intervene in—raising questions of safety, accountability, and transparency.
Using regulatory insights from HAVTIL and human factors research, Meloe mapped out a framework for responsible AI design—one that includes clear communication of system limitations, risk-informed design of human-machine interfaces, and the ability to monitor and override automated actions when necessary. He argued that trustworthy AI requires more than clean code or clever algorithms—it demands high-quality data, domain-specific knowledge, and an engineering mindset focused on reducing uncertainty and error.
The session also showcased DNV’s practical work in integrating knowledge into AI systems—from converting unstructured inspection reports into structured, contextualised data, to building graph-based models that enhance explainability and auditability across industrial applications.
Ultimately, Meloe’s message was clear: the future of AI in critical industries depends on our ability to combine technological innovation with robust governance, human-centric design, and a deep commitment to transparency and operational safety.
