Report from the Workshop on Requirements for Regulatory, Insurance and Legal Acceptance of Automated Decision-Making in NDT, SHM and CM Technologies, now published by BINDT
RCNDE was proud to play a leading role in the Workshop on Requirements for Regulatory, Insurance and Legal Acceptance of Automated Decision-Making in NDT, SHM and CM Technologies, bringing together regulators, insurers, legal experts and the wider NDE community to explore the opportunities and challenges of automated decision-making technologies.
The workshop provided a valuable platform for collaboration, helping to define the requirements and pathways needed to support the future adoption of AI-enabled and automated inspection technologies across industry. Special thanks to The British Institute of Non-Destructive Testing (BINDT) and all contributors for making this important discussion a success.
The workshop concluded that while AI-powered decision-making in non-destructive evaluation (NDE) has significant potential to improve safety, reliability and asset management, wider adoption will depend on strong governance, regulation and assurance frameworks.
Key takeaways included:
- Confidence in AI-supported decision-making increased among participants during the workshop, although a cautious and realistic approach remains necessary.
- The main barriers are not the technology itself, but the processes, evidence, governance and regulations needed to ensure trust and accountability.
- The NDE community should focus on developing frameworks that protect designers, asset owners and operators from legal, regulatory and insurance risks associated with automated decisions.
- Lower-risk, high-volume applications, such as manufacturing inspection, are expected to be early adopters of automated decision-making technologies.
- Successful implementation will require collaboration between industry, regulators, insurers and legal experts to establish clear requirements and standards.
Recommendations:
1. Messaging
- Clearly communicate what AI can and cannot do, with messaging tailored to specific use cases.
- Position AI as a decision-support tool, rather than a replacement for human decision-makers.
- Focus on AI’s role in reducing risk and preventing failures, rather than simply finding more defects.
- Emphasise the long-term lifecycle value of AI, not just inspection cost savings.
- Develop standardised business case templates that consider cost, risk, quality and availability.
2. Implementation Approaches
- Learn from AI adoption experiences in other sectors, such as healthcare, autonomous vehicles and EVs.
- Introduce AI alongside existing systems to build confidence through parallel operation.
- Adopt a phased implementation strategy, beginning with small-scale pilot projects.
- Create trial environments and regulatory sandboxes to test technologies safely.
- Establish regulator-approved pilot frameworks.
- Prioritise low-risk, high-volume applications as early use cases.
- Develop methods for sharing anonymised data and best practice where possible.
- Create AI-focused codes of practice that can evolve into formal standards.
3. Confidence Building and Assurance
- Define minimum requirements for explainability and auditability.
- Require robust evidence supporting AI decisions, even when full transparency is not possible.
- Build confidence through demonstrated reliability, availability and performance.
- Establish formal validation methodologies for AI systems.
- Engage insurers early and provide operational evidence rather than theoretical benefits.
- Encourage collaboration across the insurance sector to share learning and best practice.
- Clarify accountability and the responsibilities of the “competent person.”
- Maintain strong configuration management, documentation and digital traceability to support future legal and regulatory scrutiny.
4. Skills and Roles
- Update training programmes to include AI fundamentals, limitations, governance and oversight.
- Define new competency roles, such as AI supervisors and validators.
- Maintain practical expertise through dual-system operation and ongoing refresher training.
Overall Message
Successful adoption of AI in NDT, SHM and CM will depend not only on technological capability but also on trust, governance, assurance, regulation, skills development and clear communication. The recommended approach is a gradual, evidence-based deployment that keeps humans accountable while building confidence among regulators, insurers, operators and asset owners.
