How can we apply the defence in depth approach to preparing for risks from AI?
Background
In the National AI Strategy, the government made commitments to enrich our understanding of AI as it impacts the economy and society more broadly. Additionally, we recently launched a steering board chaired by the heads of both the government analysis and scientific functions, to ensure cohesive cross government approaches to understanding AI impacts. An overview of the high-level questions we are asking in this regard are outlined in the section below. (https://www.gov.uk/government/publications/national-aistrategy)
Some priority work we are currently developing to meet these commitments include:
An analysis of the AI White Paper consultation to feed into the formal consultation response. This will allow us to take on feedback from the public and various key players in sectors across the economy, and better tailor policy interventions to support strategic AI aims.
Establishing the AI Safety Institute to advance the world’s knowledge of AI safety by carefully examining, evaluating, and testing new frontier AI systems. The Institute will conduct fundamental research on how to keep people safe in the face of fast and unpredictable progress in AI, improving our understanding of the capabilities and risks of AI systems.
A monitoring and evaluation framework for AI regulatory interventions in tandem with the AI regulatory white paper. This will develop our understanding of key metrics to monitor with regards to Ai governance and ecosystem impacts.
Research into the AI sector and supply. Updating the AI Sector Study to establish a consistent and comparable set of economic indicators for the AI sector in terms of producers and suppliers. This study helps us to best understand where the AI sector needs support, to grow sovereign capability of the UK in AI, in alignment with strategic priorities.
The development of a cross-economy national AI risk register. Developed in tandem with a responsibility register that garnered cross Whitehall agreement on which departments hold which risks with regards to AI. The risk register will provide a single source of truth on AI risks which regulators, government departments, and external groups can use to prioritise further action.
Further research into Compute and the best ways to leverage compute to support the AI sector. This will be key to informing our response to the future of compute review, and maximising the £1 billion+ investments in state-of-the-art compute.
Next steps
If you are keen to register your interest in working and connecting with DSIT Digital Technology and Telecoms Group and/or submitting evidence, then please complete the DSIT-ARI Evidence survey - https://dsit.qualtrics.com/jfe/form/SV_cDfmK2OukVAnirs.
Please view full details: https://www.gov.uk/government/publications/department-for-science-innovation-and-technology-areas-of-research-interest/dsit-areas-of-research-interest-2024
Related UKRI funded projects
-
Ethical AI Marketplace-as-a-Service to help businesses impacted by COVID build back better
ProtectBox have developed an "artificial intelligence (AI) Engine" (a highly adaptable AI Marketplace-as-a-Service) with many user-friendly features that lets B2B/C/G/I users (in minutes, for free) assess, matc...
Funded by: Innovate UK
Why might this be relevant?
The project focuses on AI safety and understanding AI capabilities and risks, which is relevant to preparing for risks from AI.
-
ENFIELD: European Lighthouse to Manifest Trustworthy and Green AI
ENFIELD will create a unique European Centre of Excellence that excels the fundamental research in the scientific pillars of Adaptive, Green, Human-Centric, and Trustworthy AI that are new, strategic and of paramount imp...
Funded by: Horizon Europe Guarantee
Why might this be relevant?
The project aims to advance research in trustworthy AI, which is relevant to preparing for risks from AI.
-
Turing AI Fellowship: Adaptive, Robust, and Resilient AI Systems for the FuturE
The Office for Artificial Intelligence (AI) estimates that AI could add £232 billion to the UK economy by 2030, increasing productivity in some industries by 30%. However, to be truly transformational, the integrat...
Funded by: EPSRC
Why might this be relevant?
The project specifically addresses the need for a new theory and framework to enable the creation of adaptive, resilient, and robust AI systems, directly relevant to preparing for risks from AI.