AI Charter

Charter on the Responsible Use of Artificial Intelligence in Research

By: Dr. Adolphus Lye 

Risk, Reliability and Uncertainty Quantification (R2UQ)

Preamble

Artificial Intelligence (AI) is transforming scientific research, creating unprecedented opportunities to advance data-driven discovery, computational modelling, and engineering decision-making. In Risk, Reliability and Uncertainty Quantification (R2UQ), AI offers significant potential to enhance our understanding and management of risks in complex engineering systems. Conversely, the increasing deployment of AI introduces new challenges concerning reliability, uncertainty, robustness, cybersecurity, and safety, reinforcing the relevance of R2UQ in developing trustworthy AI-enabled technologies.

I recognise AI as an increasingly integral part of scientific research. Nevertheless, its adoption must remain grounded in scientific rigour, intellectual independence, purposeful application, and ethical responsibility. Guided by the philosophy of "Trust but Verify", I affirm my commitment to the following four pillars.

Pillar I — Purposeful Integration: AI for R2UQ

AI shall serve as an enabler of scientific discovery, not an end in itself. Its integration into research workflows, including machine learning, data integration, physics-enhanced modelling, and uncertainty quantification, must be guided by clearly defined research objectives and demonstrable scientific value. AI should be adopted where it meaningfully enhances research capabilities, rather than merely for novelty, convenience, or appearance.

Pillar II — Advancing Trustworthy Systems: R2UQ for AI

As AI becomes increasingly embedded in complex and safety-critical engineering systems, R2UQ has an essential role in evaluating and assuring their reliability, robustness, and safety. I commit to advancing research methodologies that address uncertainties, system vulnerabilities, cybersecurity threats, and emerging risks associated with AI-enabled technologies. The responsible advancement of AI must be accompanied by equally rigorous efforts to understand, quantify, and manage its risks.

Pillar III — Intellectual Independence: Trust but Verify

AI shall augment, never replace, independent scientific reasoning and critical judgement. AI-generated analyses, computational results, interpretations, and research narratives must be critically evaluated through appropriate verification, validation, and physical consistency checks. Research findings should remain scientifically defensible, reproducible where feasible, and transparent about material AI involvement. The ultimate responsibility for scientific integrity, intellectual contributions, and research conclusions rests with the human researcher and must never be outsourced to AI.

Pillar IV — Ethical Stewardship: Safeguarding Research Integrity

AI must be employed in ways that uphold scientific integrity, public safety, and societal trust. I reject its misuse for malicious cyber activities, unauthorised vulnerability exposure, fabrication or manipulation of research findings, paper-mill practices, and dissemination of misinformation. Responsible AI adoption also requires appropriate safeguards for research data, transparency, accountability, and responsible disclosure. While effective AI governance remains an evolving challenge, I commit to supporting proportionate safeguards and continually reviewing research practices as technologies, risks, and ethical standards evolve.

Closing Commitment

I believe that the responsible integration of AI requires a deliberate balance between technological innovation and scientific accountability. Through this Charter, I commit to harnessing AI to advance R2UQ research while contributing to the development of safer, more reliable, and trustworthy AI-enabled engineering systems. Above all, AI must remain a tool that strengthens human scientific intelligence, rather than a substitute for human understanding, judgement, and responsibility.

"As we develop AI to advance engineering research, we must equally advance the science needed to understand and manage the risks of AI itself." 

— Dr. Adolphus Lye