This digest features the latest insights from the AI Exchange, a partnership between CAST and IVAR, set up to help charities and foundations engage with AI in ways that are responsible, equitable and rooted in trust.
What we’re learning about AI in the voluntary sector
Charities are harnessing their AI skeptics
Some small charity leaders - stretched already to capacity - are understandably reluctant to navigate what AI means for their work and their team but recognise that they need to address it. To overcome this challenge, some leaders have delegated AI to other team members or nominated ‘AI champions’ to pick up this mantle. For the leader it’s one less thing to carry, while ensuring the organisation continues to consider how AI can affect their organisation. (We have a new resource specifically for charity leaders exploring AI.)
What about teams’ AI cynics? We’re heard that leaders who are more active in their use of AI are drawing on more skeptical colleagues to sense-check their use of AI and ensure it’s proportionate and appropriate. This helps to provide a check and balance and a ‘human in the loop’ as well as recognising that it can be useful to have team members with varying levels of comfort using AI - not everyone has to be at the same level for an organisation to come up with a considered approach.
Charity leaders continue to emphasise the importance of making space to try AI safely and proportionately. Experimenting and testing is often the best way to understand how AI can be used so setting up processes and tools that enable ‘safe failure’ can help the organisation to move forward responsibly.
These insights come from IVAR's Leading in Uncertainty and AI sessions that form part of the AI Exchange.
AI in funding applications: Where it’s useful
We’ve seen a dramatic increase in the number of funders using AI in funding application processes. At a recent workshop, funders shared some of their use cases including:
- Summarising applications for panels and committees to make content more readable and reduce the volume of materials.
- Early-stage screening for eligibility and remit against listed application criteria.
- Due diligence including reviewing charity accounts and annual reports.
- Drafting feedback.
- Consistency checking: comparing answers and scores when an organisation reapplies across different funding rounds.
- Flagging AI use in funding applications.
- Data analysis and visualisation: one funder shared their use of Claude to build dashboards analysing ward-level data.
Although this list may suggest that much is being replaced by AI, for most funders, AI is being used more as an assistant. One noted:
‘Everything it does is not replacing a human, it’s just getting AI to do that work at an earlier stage to help identify questions and challenges at an earlier point’
Funders identified a number of opportunities that arise from using AI in these ways:
- Creating efficiencies – several participants were surprised about how much time AI can save when used well.
- Enabling new practices, including ones that may increase accessibility for both applicants and assessors (e.g. voice-to-text AI submissions).
- Better strategic data, including potential for monitoring around equity, diversity and inclusion.
- Scope for more funder collaboration: sharing information and trusting each other’s due diligence, by providing rigour and shared process behind more joined-up work.
These insights come from our bridging workshops as part of the AI Exchange, as well as the AI for Grantmakers Peer Group.
AI errors
We’re keen to hear about the AI mistakes you’re seeing as you implement AI in grant-making processes. What is AI getting wrong and how are you managing these errors? If you have an instance (or many) that you’d like to share with us, we’ve set up a fully anonymous one-question form to capture your experiences of AI errors. Open to anyone, with no deadline.
Upcoming events
AI makes mistakes: A workshop exploring how funders manage AI errors
AI’s confidence can lull us into a false security that everything it does is correct. This makes its errors liable to pass us by - and trip us up. This is a workshop for funders to share their experiences of AI getting it wrong and to help shape good practice for when that happens. The workshop aims to be a space to share experiences, examine what went wrong and how to fix it, and consider where collaboration might support solutions and the sector. We plan to produce an anonymised output for our communities navigating AI. Register to attend here.
What we’ve been busy with at the AI Exchange
Bridging workshops
Grant applications
In July, we convened a small group of funders who are incorporating AI in their funding application processes. The workshop highlighted that funders are using AI across a range of application stages and in some cases, it is enabling funders to adopt new practices such as providing feedback to applicants (where they previously didn’t have the capacity to do so).
Leading in Uncertainty sessions
We brought together eight smaller charity leaders to share their reflections and responses to AI in July. The aim is to provide peer spaces for open and honest reflection on the tensions, opportunities, and impacts of AI use without pressure to adopt or implement, as well as tips and advice.
Grantmaking in the Age of AI: Keeping Trust
On September 10th, CAST, IVAR and The Technology Association for Grantmakers, brought together nearly 30 practitioners, funders and charities for a one-day design studio to brainstorm ideas for building and strengthening trust in grant-making. Three key ideas emerged: a secure charity passport for applicants to be able to share information easily with funders; a discussion forum for funders and charities to support honest reflections; and a variety of mechanisms for surfacing hidden gems - the small grassroots organisations that often not even applying for funding. We’ll be exploring these ideas further before the end of the year.
AI in our organisations
At IVAR, since developing our AI principles we’re adding two supporting documents. The first is a policy that puts our principles into action and outlines specific practice across our activities. Our second is a strategy - a guide for where we think AI adds value to our work and where we think AI has the potential to undermine our contributions. These documents wouldn’t be possible to draft without an initial period of experimentation and testing to explore the possibilities - this has been an action learning journey for us.
At CAST, we're exploring more intentionally how to use AI as a team, rather than as a group of individuals. Moving from siloed to more collective use of AI is a challenge for many charities, and we're no exception, but it felt like the natural next step. Building on the insights from our previous 'AI Together' sessions, we're now trialling Claude Team to work more consistently and collaboratively on shared use cases. Through monthly sessions, we've been trying out its collaborative features, like shared projects and a shared design system based on our brand guidelines, to help the team build confidence with some of our more common use cases.
Resources
The humans behind the AI Exchange
I'm David, Programme & Partnerships Lead at CAST, where I lead the Digital Leads Network and some of our Worthwhile AI work. My background is in digital inclusion and social innovation, so I'm most interested in exploring how AI can work for the communities that are usually not at the forefront of AI development. I use AI for both work and personal life, from drafting and planning to building small, helpful tools like a Spanish conversation buddy, and taming endless school admin for my children! I share bite-sized experiments in my Lo-Fi AI series. I'm both curious and wary about where this is all heading, so I like to try small, low-risk experiments to get a feel for new tools, and keep a critical eye on the latest developments.