Centre for Data Ethics and Innovation Minutes - Advisory Board Meeting 8
Updated 12 October 2023
© Crown copyright 2023
This publication is licensed under the terms of the Open Government Licence v3.0 except where otherwise stated. To view this licence, visit nationalarchives.gov.uk/doc/open-government-licence/version/3 or write to the Information Policy Team, The National Archives, Kew, London TW9 4DU, or email: psi@nationalarchives.gov.uk.
Where we have identified any third party copyright information you will need to obtain permission from the copyright holders concerned.
This publication is available at https://www.gov.uk/government/publications/board-meeting-agendas-and-minutes/centre-for-data-ethics-and-innovation-minutes-advisory-board-meeting-8
5th September 2023
14:00-15:30
Board Attendees:
- Prof Neil Lawrence - Interim Chair
- Dr Rumman Chowdhury
- Martin Hosken
- Dr Marion Oswald
- James Plunkett
- Eddie Copeland
- Dr Mimi Zou
Apologies:
- Jack Clark
CDEI Attendees:
- Felicity Burch
- Louise Sheridan
- Team Leads
- Secretariat
Guest attendees (item 4):
- Rt Hon Sir John Whittingdale OBE MP
Agenda
Item 1: Welcome
NL welcomed attendees, noting that this is the last Advisory Board meeting. He invited FB to take the floor.
FB thanked the Board Members for their commitment and contributions during their term, highlighting that the CDEI team is keen to keep Members involved and engaged with their work, and will be flagging opportunities for Members to do so even after their terms end.
Item 2: CDEI portfolio updates
NL invited the team to provide updates against their programmes and the board to offer feedback and advice.
0.1 CDEI support on government’s work on AI Safety
The team explained that given increased emphasis on AI safety across government, CDEI’s expertise has been called upon in a number of areas, including:
- Leading the development of the Government’s response to the House of Lords Inquiry into Large Language Models.
- Leading the design and delivery of the Manchester Prize - a £1m annual prize for 10 years, announced by the Chancellor at 2023 Budget.
- Supporting the development of the UK’s leadership in AI Safety and engagement through the AI Summit.
- Supporting on AI skills to ensure the UK is well positioned to attract and grow a world beating AI workforce.
- Supporting the development of the government response to the independent ‘Future of Compute Review’ published in March 2023.
In response the Board:
- Highlighted the potential ambiguity of the words ‘AI safety’ and importance of having a clear definition in the context of the AI Safety Summit. The CDEI team noted that for the purposes of the Frontier AI Taskforce and the AI Safety Summit, the word ‘safety’ refers to long-term frontier risks concerning AI.
0.2 AI Assurance
The team provided an update on the programme, in particular:
- Confirming the launch of the Fairness Innovation Challenge on 16 October, with an in-person event on 19 October.
- Developing the methodology to create a taxonomy/ontology of AI Assurance concepts.
- Beginning the development of two assurance tools - AI Management Essentials, and AI Product Essentials.
- Publishing the second iteration of the Portfolio of AI Assurance Techniques in September 2023. CDEI case studies will also be uploaded in the OECD Catalogue of tools and metrics.
- Developing an accessible publication aimed at creating a simplified introduction to AI Assurance.
- Updating guidance on procurement, development, and deployment of AI in HR & Recruitment.
In response the Board:
- Praised the AI Assurance work and the positive international response to the programme.
- Encouraged the AI assurance team to engage with the US National Institute of Standards and Technology (NIST) and Responsible AI UK if not already doing so.
0.3 Responsible Data Access
The team provided an update on the programme, in particular:
- Publishing the “Enabling responsible access to demographic data to make AI systems fairer” report which informs the development of the Fairness Innovation Challenge.
- Scoping potential PETs pilots delivered in collaboration with the US, and developing a joint blog series to highlight lessons learned from the PETs Prize Challenges.
- Progressing a PETs cost-benefit analysis tool in collaboration with the ICO, following the tool’s announcement in June alongside the publication of the ICO’s PETs guidance.
- Developing a cross-government PETs community of practice, in collaboration with the ONS’s Data Science Campus.
0.4 Algorithmic Transparency
The team provided an update on the programme, in particular:
- Working on the modularisation update for the Algorithmic Transparency Recording Standard (ATRS); the design of this update will be informed by cross-cutting Model Card work in order to harmonise ATRS with procurement products.
- Continuing stakeholder engagement on the ATRS, noting particular engagement with local government; the team also highlighted the translation of the ATRS into Welsh.
In response the Board:
- Flagged the importance of the Algorithmic Transparency work and the positive international response for the programme.
- Expressed concern about team resourcing given the ongoing AI Summit and Taskforce surge, arguing for a need to keep up momentum on algorithmic transparency.
0.5 Strategic Projects
The team gave an overview of the work being carried out in partnership with cross-government teams, and updated that the work on redesigning the CDEI Model for Responsible Innovation is now complete.
In response the Board:
- Advised on how to encourage engagement for the Model for Responsible Innovation with new audiences, noting that there is a proliferation of guidance at the moment and that it would be useful to have it housed in one place on GOV.UK for example; the CDEI team agreed with this assessment and mentioned that our ongoing collaboration with CDDO’s landscape review seeks to identify and align cross-Government guidance. CDEI is also taking a more product-focused approach across its programmes as part of the modularisation work that will help facilitate user understanding of available guidance.
Item 3: Project deepdive
The CDEI team presented on the progress of the Fairness Innovation Challenge, an end-to-end grant challenge that will support the development of innovative solutions to address bias and discrimination in AI systems. The presentation focused on the use cases and evaluation criteria being considered as well as the engagement and communications approach ahead of the launch of the Challenge.
In response the Board:
- Highlighted legal/courts and recruitment use cases as being of particular interest.
Item 4: Update from Rt Hon Sir John Whittingdale OBE MP
NL welcomed Minister Whittingdale to the meeting. The Minister thanked the Advisory Board Members for their commitment and contributions during their term, noting that going forward the Department would be keen to draw on their expertise even after the end of their term.
In response the Board:
- Praised the work of the CDEI and the breadth of the team’s expertise, as well as the successful ways of working between the team and the Board.
- Raised concerns around the current framing of the AI Safety Summit and the potential for a divisive response to this framing from the wider ecosystem, and suggested that Government could look to define the term ‘safety’ in the context of AI to reduce ambiguity around the term.
- Flagged the importance of engaging with local government around the AI Summit.
- Raised concerns about how the work of the CDEI will continue to be supported amidst competing priorities on AI within the Department.
The Minister thanked the Members for their feedback and reiterated his thanks for their work on the Advisory Board.
Item 5: Meeting Close