ESRC Digital Good Network

The Public Voices in AI Fund

Pvai Fund

The Public Voices in AI Fund invites proposals for projects which seek to ensure that uses of AI are informed by the voices of people underrepresented in or negatively impacted by AI. By ‘uses of AI,’ we mean AI research, development, deployment and policy.

We invite project proposals up to the value of £50,000 which are led by organisations in the Voluntary, Community and Social Enterprise Sector (VCSE).

Background

The Public Voices in AI Fund is part of Public Voices in AI, which aims to ensure that public views and voices are front and centre in all uses of AI. Public Voices in AI builds, shares and reviews knowledge and understanding about public views on AI and encourages a range of AI users and stakeholders to engage diverse communities in AI uses, research, development, use and policy-making.

Public Voices in AI is funded by RAI UK (which is funded by UK Research and Innovation (UKRI)). It is a collaboration between the Digital Good Network at the University of Sheffield, the Ada Lovelace Institute, the Alan Turing Institute and University College London. The project web page explains our work in more detail.

Public Voices in AI runs from April 2024 to March 2025. Because of this short timeframe , the Public Voices in AI Fund a) requires applications to build on established relationships with groups negatively affected by or underrepresented in AI and b) invites proposals that build on existing activities.  Projects funded by the Public Voices in AI Fund must be complete by the end of 2024 with a final project output due by 31 January 2025.

Different groups and people benefit from and are affected by AI differently. From facial recognition technology to large language models, AI is having particular impacts on minority and already-disadvantaged groups, for example in its application in decision-making in health, immigration, social security and justice contexts. Groups which are most affected often have least resources or access to power to shape AI technologies and to participate in shaping AI. The Public Voices in AI Fund therefore centres on those groups and people which are most impacted and underrepresented.

What we’re looking for

We invite proposals for funding of up to £50,000 for projects which seek to ensure that uses of AI are informed by the voices of communities underrepresented in or negatively impacted by AI.

We want to support projects that are developing new ways of engaging communities in dialogue about how AI is used and implemented.

We will support projects which address the impacts of AI on underrepresented groups or which identify ways in which AI could be used to reduce discrimination and bias.

Proposals should be led by organisations in the VCSE Sector that are led by or work with groups and people negatively affected by or underrepresented in AI. These include, but are not limited to: racialised minorities; low-income groups; LGBTQ+ groups; women; people with disabilities; people dependent on public sector support (eg Universal Credit); migrants, refugees and asylum seekers; people with experience of homelessness, clinical, financial or other vulnerabilities. We welcome proposals that explore multiple disadvantages.

Proposed activities must:

  • be a form of participation or participatory research that goes beyond merely consulting towards involving and empowering groups or communities so that they are able to effect change;
  • build on existing work and/or relationships with underrepresented communities;
  • aim to address unequal power dynamics between underrepresented groups and AI decision-makers;
  • be undertaken in ways that foster an equitable and ethical participatory environment (for example, fair and proportionate pay for participants’ time, a commitment to providing a return to the communities engaged, a trauma informed (if appropriate) and/or non-extractive approach).

Proposals can build on existing activities which are already being undertaken. For example, proposals may seek to improve methods already in use, to scale up activity already underway, to support communications and dissemination of existing activity, or to amplify impact.

Who is eligible to apply?

Applications must be submitted by one lead applicant and can include multiple team members.

We invite applications from lead applicants in the VCSE Sector. The lead applicant’s organisation should be either:

  • a charity, civil society or community group working directly with participants who have lived experience of under-representation in or negative outcomes from AI,

OR

  • a social impact and social enterprise organisation. We are particularly interested in supporting applications from organisations that are led by the communities they seek to serve.

The lead applicant’s organisation must have financial and administrative capacity to manage the grant, including a company bank account. Awards will be conditional upon passing due diligence checks, explained below.

Applicants should demonstrate that they have established relationships with groups negatively affected by or underrepresented in AI.

All team members and projects must be UK based.

Funding available

We will make a maximum of £170,000 available to fund a range of projects. We envisage funding 4-6 projects of up to £50,000. UK business, third sector or government organisations can claim costs at 100% of eligible direct costs to the organisations. According to UKRI criteria, eligible costs vary slightly depending on the organisation type:

 a) Business
b) Third sector in receipt of a subsidy
(in line with the UK Subsidy Control Act 2022)
Third sectorGovernment
Staff – salaryYesYesNo
Staff – NI/SuperannNoYesNo
Total Research Staff – FTE limits1 year FTENo FTE limitNo
Travel and subsistenceYesYesYes
Other direct costsNoYesNo
OverheadsNoYesNo

Other organisations such as Universities can claim costs at the standard UKRI rate of 80% of the Full Economic Cost (fEC). Information about such organisations can be found on UKRI’s eligibility as an organisation webpage.

All costs are subject to UKRI funding guidance

Examples of common eligible costs are given below:

  • working time for lead applicant/team members (eg to run workshops and recruit participants)
  • participant honoraria and expenses
  • materials (eg artistic) to support the participatory work
  • travel and subsistence appropriate to delivery of the project
  • events related costs, eg venue hire.

This is an indicative list of eligible costs – it is not exhaustive. In line with UKRI regulations, alcohol is not a permitted cost.

If you have any questions about costing eligibility, please email PublicVoicesinAI@digitalgood.net.

Funding logistics

Funds are usually paid in arrears, but some upfront payment can be provided where it is required.

How we will assess your application

Submitted applications will be checked against the eligibility criteria (listed under ‘Who is eligible to apply?’ above) and any applications that are clearly not eligible will not proceed to review.

If we receive a high volume of applications, applications may be screened by a member of our team to ensure that they meet the scope of the call (as described in ‘what we’re looking for’ above). Applications that fall outside the scope of the call will not proceed to review.

Eligible applications will then be reviewed by two members of the Public Voices in AI team. Principles of peer review, including relating to conflicts of interest, will be strictly adhered to.  As the Public Voices in AI Fund is led by the Digital Good Network, it follows their policy on conflicts of interest.

Reviewer scores will then be collated, applications will be ranked and any discrepancies will be discussed. A panel comprising at least one member of the Public Voices in AI team and at least one member of our Steering Group will make a final decision on which proposals to fund. This decision will be based on a) this ranking, b) diversity of represented communities, region and geography and c) meeting the priorities of the funding call.

Assessment Criteria

Criteria below are equally weighted.

Engagement of impacted or underrepresented communities

This includes whether:

  • proposals engage groups or communities underrepresented in  artificial intelligence research, development and policy (AI RD&P), including but not limited to: racialised minorities; low-income groups; LGBTQ+ groups; women; people with disabilities; people dependent on public sector support (eg Universal Credit); migrants, refugees and asylum seekers; those who are homeless, people experiencing clinical or financial vulnerabilities, or vulnerabilities of any other kind. Many people may belong to different underrepresented communities, experiencing multiple disadvantage so this can also explore the intersections,
  • proposals are led by the communities they aim to serve (e.g. the lead applicant is part of an underrepresented community, and/or can demonstrate how members of the community will shape the direction of the project),
  • proposals indicate how proposed activities could address unequal power dynamics between underrepresented groups and AI decision-makers.

Methods

This includes whether:

  • proposals develop new ways of engaging diverse communities in AI RD&P,
  • proposals are tailored to the particular needs of the community, issue and context,
  • proposals ensure that communities play a role in defining the issues to be addressed.

Social value

This includes whether:

  • proposals demonstrate clear plans to empower community members to build skills and knowledge, including but not limited to skills and knowledge related to AI,
  • proposals input into the broader Public Voices in AI project, as described in ‘Conditions of Funding’ below ,
  • proposed activities produce results or outputs that will be of interest to policy-makers and/or other power-holders.

In addition, proposals will be assessed against the following standard UKRI assessment criteria:

Excellence and impact

This includes:

  • whether the proposal will generate new knowledge and lead to new insights,
  • clearly described and justified approach and design,
  • clearly articulated plans for working ethically,
  • if data is gathered, clearly described and justified data management and access plans that identify the risks to, and mitigations for, accessing, managing and sharing data.

Value for money

This includes whether:

  • the funds requested are essential for the work,
  • the importance and potential justify funding on the scale requested,
  • there’s a clear allocation of duties and responsibilities,
  • the proposal represents good value for money.

Conditions of funding

It is a condition of funding that lead applicants commit to a member of their team contributing to the Public Voices in AI project by:

  • engaging with Public Voices in AI team members to contribute to the development of project resources, to be produced and shared with a range of AI stakeholders (likely to be 2 x one-hour interviews),
  • attending two events: a) a one-day meeting to contribute to Public Voices in AI resources; and b) a final project event in early 2025 (before the end of March).

Please include in the budget costs to cover two days of time (for one person) to enable project teams to satisfy the above conditions of funding.  Travel and subsistence costs of attending these events will be covered by the project and does not need to be included in the project budget. 

We also require that successful lead applicants:

  • Engage in our due diligence processes before awards can be made.  This means that the Due Diligence team at the University of Sheffield will carry out some desk based research on the lead applicant’s organisation relating to ethical, legal, financial and national security considerations.  They may also contact applicants to ask for accounts information and a (limited) amount of other information.
  • Sign a funding agreement with the University of Sheffield – a copy can be viewed here.
  • Produce a final project output within one month after project completion (format to be agreed).

Timeline

We recognise that our timeline is tight. This is because the funding that we have received must be spent by the end of March 2025, a timeline which is out of our control. It is because of this timeline that a) we ask applicants to demonstrate that they have established relationships with groups negatively affected by or underrepresented in AI and b) we invite proposals that build on existing activities.

  • Call launch: Friday 31 May
  • Deadline for submissions: Wednesday 3 July 5pm (UK time)
  • Decisions communicated to applicants: by end July
  • Projects cannot start before: 1 September 2024
  • Projects must be complete by: 31 December 2024*
  • Final project outputs to be received by: 31 January 2025
  • Public Voices in AI final event: TBC (will be early 2025 – before end of March)

* Due to the short duration of Public Voices in AI, extensions beyond December 2024 will not be possible.

Support for applicants

Join our online Public Voices in AI launch event: Thursday 6 June 1pm (UK time). 
Sign up here.

Please see our FAQs below.  If you have a question that isn’t answered in our FAQs, please contact the team on PublicVoicesinAI@digitalgood.net  Please note that we are not able to provide feedback on individual applications before submission.  All enquiries will be answered in a similar way, so as not to advantage one applicant over another.

How to apply

Please download our application form here. If you need an alternative format please email PublicVoicesinAI@digitalgood.net to discuss your requirements.

Please email your application and attachments in one .pdf document to PublicVoicesinAI@digitalgood.net by 5pm (UK time) on Wednesday 3 July 2024.

Supporting documentation

  • We require a one page CV from the lead applicant.
  • We require a letter from all organisations involved confirming their support for the project.

Please consult the application form for full details.

FAQs

I don’t have a PhD – can I still apply?

Yes – we do not require team members to have a PhD.

What does full economic cost (fEC) mean?

The full economic cost of a project is the full cost to the organisation of undertaking the activity.  For certain organisations UKRI will only give out funding at 80% of the full economic cost and the organisation must agree to find the remaining 20% from other resources.  This, however, only applies to organisations that already meet UKRI eligibility criteria – for example universities.  UK business, third sector or government organisations can typically claim costs at 100% of the direct costs to the organisations.

Further guidance on existing eligible UKRI organisations can be found here – Eligibility as an organisation – UKRI.

Can the lead applicant be a University?

No.  The lead applicant’s organisation must be a either A) a charity, civil society or community group working directly with participants who have lived experience of under-representation in, or negative outcomes from, AI OR B) a social impact and social enterprise organisation (we’re particularly interested in organisations that are led by the communities they seek to serve (e.g. the lead applicant is part of an underrepresented community, and/or can demonstrate how members of the community will shape the direction of the project)).

However, we welcome applications from teams that include University based members.

Can a self employed person apply?

Self employed people cannot be the lead applicant but they can be part of a project team.

Are you able to offer extensions to the timeline?

No – unfortunately the timeline set by our funder is very tight (our project only lasts 12 months) and we therefore have no scope to be flexible with our timeline.

We are an organisation that’s based in the UK but we work with groups who are based outside of the UK – can we apply?

No – this call is only open to teams and projects that are based in the UK. We would encourage you to consider the Digital Good Network’s Digital Good Research Fund if you are an international team: https://digitalgood.net/opportunities/digital-good-research-fund/. The next call is expected in early 2025.