This funding opportunity aims to support projects that address strategic themes that underpin responsible AI.
It aims to grow the network of researchers engaged in these topics and complement:
- other key aligned programmes already funded by UKRI (for example, The Alan Turing Institute, Ada Lovelace Institute, AI hubs, Centres for Doctoral Training)
- other institutions in the UK considering the development and deployment of AI technologies
Keystone projects will form core research pillars of RAI UK which seeks to bring together the best talent to address multidisciplinary and interdisciplinary challenges posed by responsible AI.
We particularly encourage applications that fit the RAI UK vision, which is to:
- help enable responsible AI to power benefits for everyday life
- work in collaboration with researchers, industry professionals, policymakers and stakeholders to drive an ecosystem that will be responsive to the needs of society
Applications must clearly articulate how they address at least one of the following themes. These themes have been defined through:
- a range of consultations including the series of roundtables that have been run at townhall events in London, Cardiff, Belfast, and Glasgow
- engagements with the AI ecosystem through a series of roundtables
- feedback from members of the RAI UK Strategy Group and other key members of the UK AI community
Further detail on the themes and the questions they wish to address:
Responsible AI-powered organisations and economies
Organisations and economies are rapidly evolving due to new ways of working in response to AI. This has wide-ranging implications for organisations and their employees, the end-users of the products and services offered by these organisations, and their economies.
Responsible AI practices need to include systems-based approaches, beyond testing of individual AI algorithms, to evaluate the wider implications of AI deployment into complex human and AI systems.
This includes ensuring that innovators understand the overall consequences and measures, such as the reskilling and upskilling that will need to be developed. What are the responsible AI principles appropriate for sectors (for example, healthcare and the justice system) and contexts (for example, where there may be high consequences of failure) so that AI innovations are safe, trustworthy, and provide confidence to organisations, customers, stakeholders, economies, and the wider society?
Addressing harms, maximising benefits of AI
How AI is framed and applied introduces new opportunities but also trade-offs for individuals, industries, and societies where the potential benefit may be outweighed by negative impacts on a wide range of issues. This includes issues such as:
- privacy
- bias
- accessibility
- labour rights
- social justice
- sustainability (of people, organisations, and the environment)
Many of these trade-offs have direct legal ramifications.
Moreover, AI solutions often do not generalise to new settings, exacerbating uncertainties as to whether existing technical successes can translate to new domains, sectors, cultural and global contexts.
This introduces the need for deployment, validation, provenance, and auditing regimes for AI. This is so decision makers can thoroughly understand and manage the limitations of AI systems to ensure they are safe, ethical, and beneficial, while simultaneously highlighting where further development is needed to extend their scope.
Dimensions of AI safety (for example, ensuring the system functions as intended with regards to ethics, policy, legal and technical aspects), AI security (for example, ensuring the system is robust to malicious interference) and the countering of AI misuse offer may open questions to be addressed.
What is needed to be in place so that AI works for the benefit of people and societies while harms are minimised?
AI law, governance, and regulation in an international context
The UK needs to meet the challenges of when and how to govern and regulate AI within the international digital economy.
In the National AI Strategy, the UK government emphasised the desire to encourage startups and small and medium-sized enterprises to adopt AI, while acknowledging the increasing need to meet AI regulation.
However, there are few established routes to control, transparency and redress for users in relation to automation in digital public services, as well as crucial private sector interactions. For example, the rise of generative AI has posed significant (though not unique) challenges to the creative industries.
Currently very different approaches to AI governance are taken by the EU, China, the US, and elsewhere, a global debate to which the UK must make a strong contribution or risk isolation.
While longer-term AI safety issues have recently been highlighted by the UK government, short to medium term risks remain outstanding.
Against this background, we would seek the most effective research contributions that the UK can make to this global debate, to drive for clarity around AI regulation that is rights-respecting.
What are the approaches that promote trust, provide fairness, and accountability for users, and provide certainty for international commerce?
In addressing these themes, projects should consider both the technology development processes that embed responsibility and trust and how these can be shaped to deliver societal benefits through governance, regulation, and education.
Expertise
Each application should be a consortium incorporating a diverse range of voices, through engaging different academic institutions and disciplines, policymakers or advisory groups and non-academic stakeholders including commercial enterprises and civil society groups.
The team will reflect the strengths represented around the UK, nationally. It is expected this group will have:
- thematic expertise reflecting the communities that will be engaged in the investment
- in-depth understanding of relevant disciplines, technologies, policy challenges and evidence needs
- an ability to engage with diverse stakeholders, including non-academics
- an ability to articulate a clear vision for engagement with communities working in this area and new partners from within the appropriate disciplines or elsewhere
- an ability to deliver complex projects to time and on budget, considering the variety of activities and outputs
- specialist expertise (academic or non-academic) to support the desired outcomes, which may include:
- sector representatives
- programme management expertise
- a communications function
- knowledge mobilisation expertise
Key features of a keystone project
Quality and ambition
A keystone project is seen as a scheme that matches best with best and allows researchers to tackle bigger, more open-ended problems, addressed through a more coherent or holistic approach.
The stability in tackling a longer-range vision helps motivate teams, provides the freedom required to take risks, and enables longer term planning.
Partnership and ecosystem development
The scale of activity is seen to create stronger links between the universities involved and greater visibility at a national and international level, leveraging the RAI UK network, partners, and international connections.
The size of keystone projects allows for the assembly of the best team and collaborators, all with complementary expertise leading to the development of effective multidisciplinary and cross-disciplinary working. The team will be expected to promote and champion RAI UK in their respective domains, to help grow the community across sectors and disciplines.
The duration of keystone projects allows investment by the team in building effective collaborations. The scale of a keystone project should look to attract partnerships beyond the original project partners and seek greater input from the wider community, including public or private sector. This will result in more external visibility on the research direction for the area.
Keystone projects will form part the RAI UK core pillars and contribute activities that will help connect and drive efficiencies within the UK AI ecosystem. They will have access to the broad network of partners brought together by RAI UK and support international conversations on Responsible AI using the reach that RAI UK affords. They will also bring on other parts of the AI ecosystem that are currently disconnected from national conversations and research programmes. These activities should be specifically costed into the programme to allow for this flow on knowledge across organisational boundaries.
Please make it clear where the project looks to build on connections to existing networks and research programmes. RAI UK is keen to understand how this funding is building new connections and developing exiting relationships within the ecosystem, providing additional value within the landscape.
Creativity
The keystone mechanism provides freedom to scope new opportunities, allows the team to cross-fertilise ideas, and build up new skills sets. This allows the team to develop new themes, and to trade ideas and resources. The stability of the grant allows early career researchers (ECRs) in the team to express their creativity and to lead on part of the investigation.
Impact and advocacy
Keystone projects are seen to have greater visibility and recognition within the universities involved and the relevant research communities at both a national and international level, leveraging the RAI UK brand. This gives the keystone project team more influence than smaller scale research activities.
They are able to attract more visits and engagement with high quality researchers and external stakeholders, leverage other funding, and influence wider strategies. The visibility also enhances the opportunities for outreach and advocacy, promoting UK science. We expect you to demonstrate how you will deliver or support the desired outcomes in your application to support the UK’s transition to an AI-enabled economy, resulting in growth, prosperity, and mutual benefit for sectors and citizens.
Career development
Keystone projects will be a good environment for ECRs’ longer term career development. The flexibility and longer durations allow the project lead to empower junior team members giving them greater independence through more responsibility and leadership over activities.
Postdoctoral staff gain a broader experience due to the breadth of experience and expertise in the team and there are greater opportunities for secondments, mentoring and involvement in management. This makes keystone projects an attractive employment prospect leading to higher quality recruitment. PhD students would also be expected to be aligned to keystone project teams, also benefiting from interacting with a team of broader expertise and activity.
Flexibility
The flexibility keystone project holders are afforded is seen as a real strength of the scheme. The flexibility enables a more dynamic allocation of resources and a nimble approach to recruitment, and definition of the individual projects being undertaken. Specifically, integrative activities across the RAI UK ecosystem are expected, and RAI UK will be organising a series of integrator workshops to facilitate this. We require 20% of funding and researcher time to be reserved for such dynamically defined activities to be undertaken in collaboration with:
- RAI UK
- other UKRI AI investments
- other Keystone Projects
- the new AI hubs and Centres for Doctoral Training
In addition to collaborating with the wider programme, it is anticipated that your successful team will also engage with stakeholders and users of the research. They are essential to the design, conduct and impact of application-orientated research.
It is also expected that keystone projects demonstrate strong institutional support. Letters of support from participating organisations (as a minimum from the lead institution) should demonstrate alignment to the organisations’ strategy and ambitions and indicate contributions to the project. While substantial contributions (cash or in-kind) are expected from project partners and participating organisations, there is no minimum requirement, and partnerships will be assessed on their relevance and alignment to the programme.
Note that all applications submitted will be assessed equally, irrespective of which theme, or themes, the application aligns with. However, a balanced portfolio of projects across disciplines, sectors and themes will be funded.
If you are planning to involve international partners, you will need to complete RAI UK’s trusted research section and the checklist for academia by the Centre for the Protection of National Infrastructure and National Cyber Security Centre.
Note that based on the answers to this checklist, you may need to escalate this within your institution or department for a decision.
Management and monitoring
Keystone projects should have effective management and monitoring arrangements for the investment. This should include a risk management strategy and a strategy for how the flexibility of resources will be managed.
RAI UK expects all keystone projects to establish and run an independent advisory board that will include at least one member of the RAI UK leadership team The board will provide advice and recommendations on the strategic scientific and research direction and activities (such as impact, advocacy and outreach) of the programme grant.
This independent advisory board must meet at least annually. This group should have at least 50% independent membership and an independent chair.
RAI UK strongly encourages you to consider costing in project management and other administrative support such as employing a full-time equivalent project manager, and not relying on the project lead for these duties. Projects will be able to request support from the centralised communications, networking, and event management resources offered by the RAI UK Operations team for activities that look to bring together the wider ecosystem.
Equality, diversity and inclusion
The long-term strength of the UK research base depends on harnessing all the available talent. RAI UK expects that equality and diversity is embedded at all levels and in all aspects of research practice and funding policy.
We are committed to supporting the research community, offering a range of flexible options which allow you to design a package that fits your research goals, career and personal circumstances. This includes:
- career breaks
- support for people with caring responsibilities
- flexible working
- alternative working patterns
With this in mind, we welcome applications from academics who job share, have a part-time contract, or need flexible working arrangements. Please see EPSRC’s equality and diversity web pages for further information.
Responsible innovation
RAI UK is fully committed to develop and promote responsible innovation.
Research has the ability to not only produce understanding, knowledge and value, but also unintended consequences, questions, ethical dilemmas and, at times, unexpected social transformations. We recognise that we have a duty of care to promote approaches to responsible innovation that will initiate ongoing reflection about the potential ethical and societal implications of the research that we sponsor. We encourage our research community to do likewise.
Responsible innovation creates spaces and processes to explore innovation and its consequences in an open, inclusive and timely way, going beyond consideration of ethics, public engagement, risk and regulation. Innovation is a collective responsibility, where funders, researchers, interested and affected parties, including the public, all have an important role to play. You are expected to work within the EPSRC framework for responsible innovation.
What this scheme is not for
We will not fund applications that:
- do not embed responsible research and innovation and equality, diversity and inclusion considerations into the research theme itself as well as into the research practices
- are too focused on applied research and do not demonstrate significant ambitions or risk-taking
- do not build on and complement the broader set of activities going on across the UK and international AI ecosystem
- do not have a mix of technical and non-technical disciplines engaged in joint activities
- do not demonstrate clear impact pathways
- do not demonstrate an interest in engaging with the ecosystem beyond the investigators’ own disciplines
Learn about the RAI UK programme.
Funding available
Up to £10 million funding is available through the RAI UK programme to support three to five grants for up to 47 months.
We expect to fund projects requesting funding between £2 million and £3.5 million (at 80% full economic cost (FEC)). For the avoidance of doubt, the maximum funding we will provide is £3.5 million and this is the 80% FEC figure. Standard UKRI eligibility rules apply, for details on who is eligible to receive funding, please refer to the ‘Who can apply’ section.
Please note that due to the nature of this funding, additional requirements on spending profile, reporting, monitoring and evaluation as well as grant extensions will apply. This will be reflected in the grant additional conditions, and those funded will need to comply with them. Further details are provided in the additional information section. Please note that any projects funded through this funding opportunity will have a fixed start and end date, and that no slippage to this date will be permitted.
Grants will be funded at 80% FEC. The grant can support reasonable directly incurred costs, such as research staff time, consumables, travel and subsistence, and directly allocated costs, such as investigator time, and associated overheads.
You are not required to have existing collaborations or contacts within the RAI UK programme.
Current RAI UK investigators may not lead a project, nor may their time be costed on the grant, but they may be named either as project co-lead or project advisor depending on their contribution. This does not apply to investigators on our recently funded Impact Accelerator and International Partnerships projects; however, we would ask these individuals to clearly describe the added value to the RAI UK programme that funding further time commitments brings.
Equipment
Non-capital equipment over £10,000 in value (including VAT) is only available in exceptional, well-justified circumstances. Items of equipment should be in the ‘Directly Incurred – other Costs’ heading and will need robust justification; these items will be funded at 80% FEC. Items over £10,000 will be especially scrutinised, and only permitted if its remit for this funding opportunity is clearly justified.
Note that any deviation from the spending profile beyond 5% on an annual basis is not allowed (any underspend will not be refunded, nor any overspend allowed). No-cost extensions cannot be allowed.
