The recent AI consultation by the UK Government, which ran from March 29 to June 21, 2023, offered organizations and individuals an opportunity to share their insights and suggestions regarding the UK’s proposed approach to regulating AI.
With our experience building validation tools for regulated institutions and knowledge of model risk management best practices, we at ValidMind believe we are well positioned to contribute to this critical technology shift. We gathered valuable feedback from our customers and partners, including Chief Risk Officers, model developers, and validators.
The policy paper A pro-innovation approach to AI regulation by the UK Government presents a comprehensive framework that covers five core principles that emphasize the importance of values in AI development and deployment:
- Safety, security, and robustness
- Appropriate transparency and explainability
- Accountability and governance
- Contestability and redress
In the consultation process, we highlighted several critical elements to advance trustworthy AI. Our suggestions, which aim to strengthen AI the principles previously mentioned, are tied to the following areas:
- Require organizations to follow a framework for systematically documenting and validating AI systems to help improve AI transparency. High-level standards/guidelines for documentation and testing would be determined by regulators and take into account the complexity of the system as well as the materiality of the application.
- Enable the integration of AI and Model Risk Management frameworks, such as the new SS1/23 Model Risk Management Principles for Banks, to promote a unified approach to regulating AI systems and models.
- Develop clear and concise definitions of key terms and concepts related to AI and its regulation to provide businesses with a better understanding of the regulatory landscape and to facilitate compliance with the framework.
While the deadline for initial feedback was on June 21, 2023, we encourage ongoing conversations across all AI and MRM practitioners. The collective knowledge and experiences of professionals involved in the AI and MRM sectors are incredibly valuable, but are unfortunately not widely accessible.
Sharing information among the model risk management and AI community not only shapes the future of AI in the United Kingdom but also contributes to an environment of mutual learning and growth. We believe the lessons learned, and the best practices established through this consultation, will play a key role in promoting responsible innovation in the coming years, not just in the UK but around the world.