💡 What are we building?
The Agency Fund is a multi-donor initiative investing in ideas and organizations that help people navigate toward a better future.
🙌 Highlights
The Agency Fund is launching a call for AI innovations that provide targeted mentoring, coaching, or counseling to individuals in need of support. We are particularly interested in projects that take an experimental approach to assess performance on user engagement, proximate outcomes (e.g., beliefs/knowledge), and downstream social & development outcomes (e.g., education, health or economic metrics). The deadline for submission is December 22, 2023. Learn more and apply here.
As always, here are a few highlights from our partner organizations:
- Learn how SameSame is building a mental health chatbot that automatically adapts to local context, using Reinforcement Learning (RL). This capability will accelerate their ability to respond to the challenges faced by millions of LGBTQI+ youth across the world.
- Rocket Learning, IDinsight, and Jacaranda Health are among the 15 organizations selected by google.org for support on AI for the Global Goals.
🚀 Opportunities
- The Llama Impact Grants Program offers funding and technical support to organizations using the Llama 2 large language model to address education, environment, or open innovation. The deadline to apply is November 15, 2023.
- Global Grand Challenges looks to fund proposals on equitable AI use to improve global health, with separate calls for applicants in India, Brazil, and Africa. Apply by December 6th.
- The Mental Health Funding Circle is accepting rolling applications for funding. Learn more and apply here.
Challenges
In our most recent Open Call for applications, we received many proposals that we could not consider for funding. We think we may need to communicate our evaluation criteria more consistently and transparently. Here we highlight a few proposal types that we are unlikely to consider in future:
- Projects that serve only a small community of people. We are interested in projects on a path to scale. Successful applicants will have a vision of impacting 1 million or more users by 2028. We will consider a project that targets a small geographic region or population (e.g., a single village, or fewer than 200 users) — but only if it is a “pilot” effort, and only if there are plans to expand further if effective. For small, one-time projects that are not on a path to scale, we are probably not the right funder.
- Projects that require $20,000 or less. Most of our grant-making is done through a fiscal sponsor. It is relatively costly to make small awards, so we prefer to make awards above this amount. This also aligns with our preference for larger projects that are on a path to scale.
- Projects that do not include a comparison group. We want to learn about the impacts of every project we fund. So we expect our partners to measure and report the outcomes of people who participate in the project (i.e., the users or beneficiaries). Just as important, we expect grantees to measure and report on a counterfactual: the outcomes of a randomly assigned group of people who have not participated in your program. RCTs are a common strategy for this, but they are not the only acceptable approach; we are open to A/B testing and rigorous quasi-experiments as well. We just want to know what happened as a result your program — and what would have happened in the absence of your program.
📚 What We’re Reading
This month we dug into a recent working paper from our partners at Youth Impact looking at the effects of phone call tutorials and SMS messages on the numeracy of primary school students during COVID-19 school closures. In randomized controlled trials run in five countries, phone call tutorials cost-effectively improved learning outcomes. Data from Nepal also showed that participation in the trial led to a change in the beliefs and practices of teachers.
Have an agency-related article you think we should be reading? Please send us recommendations to hawa@agency.fund.