Five-year Anniversary of Validate AI
A Big Thank You to the Validate AI Community
Over the past five years, we have worked alongside an exceptional group of partners, building a community committed to advancing trustworthy, transparent, and accountable AI. We have remained focused on advocating for the importance of validation and accountability in AI systems, contributing to the growing recognition of these critical issues in the AI assurance space. This milestone serves as a testament to the collaboration, hard work, and shared vision of those who have joined us in this journey, and it marks an exciting chapter as we continue to push the boundaries of AI assurance and validation.
The launch event at the Royal Society set the foundations for our work, bringing together leading voices from academia, government, and industry. Speakers from the ONS Data Science Campus, Google, Five AI, the University of Oxford, Imperial College London, amongst many others, helped to shape the conversation. The event provided a platform to explore and discuss key issues in AI, establishing a strong foundation for future collaborations and reinforcing the importance of validation and accountability in AI systems.
Our Validate AI conference in 2021, organised in partnership with the OR Society and other collaborators, tackled critical issues surrounding AI validation. The conference centred on the need for regulation, standards, and frameworks to ensure the validation of AI systems. It emphasised the importance of robust AI assurance systems, highlighting the transformative success of using checklists to validate AI. Speakers from Holistic AI, the Alan Turing Institute, and several leading universities contributed their insights. The event also addressed the challenge of population drift and explored solutions to this complex issue, which became especially relevant to various sectors during the COVID-19 crisis.
.jpg/:/rs=w:400,cg:true,m)
As we draw to the end of the year, it seems appropriate to look back at the progress Validate AI has made over the past 12 months. It has been an eventful year, thanks in large part to the partners from across government, industry, academia, and the third sector. We would especially like to thank Imperial College London and the Operational Research Society for their long-standing support. There is increasing recognition of the importance of the Validate AI message, with the advent of large language models and more general AI systems making society more aware of the challenges that Validate AI identified five years ago, when it launched at a meeting at the Royal Society.
Particularly noteworthy achievements over this year include:
- The AI Assurance roadshow events, with over 300 attendees hosted across various sites, launching in Edinburgh, and travelling to cities across the UK and overseas, including London and Vienna, to probe different aspects of AI Validation, including AI Procurement, Operational Research enablement of AI, and creating an AI Audit profession.
- Establishing links with overseas organisations, including European and Commonwealth countries.
- Launching discussions on the creation of an AI assurance profession, exploring what is needed for accreditation with experts from a variety of fields.
Ongoing work which should reach fruition in the coming year includes:
- Development of a Masterclass in Trustworthy AI, guiding students’ implementation of high-level trustworthy AI aspirations in a quantitative and testable way.
- A roundtable event attended by AI assurance providers to brainstorm the practicalities of launching an AI Audit profession.
- Launching the Alliance for International AI Capability Building, co-founded by Imperial College London, Vienna University of Economics and Business, and Validate AI in 2024.
- Continued development of formal tools to detect and flag poor AI performance in subpopulations.
Professor David J. Hand
Chair, Validate AI CIC
Department of Mathematics
Imperial College London