Like many experts, we think AI progress in the coming years could be transformative. Yet, the vast majority of our work to improve health and accelerate science and economic growth isn’t focused on that possibility. This RFP is a step toward addressing that gap. It will support research, policy development, implementation, and field-building to help ensure AI improves human health and wellbeing, with particular attention to risks for and benefits to the global poor.
Examples of projects we might want to support include: research on how global health grantmaking should respond to advances in AI, a technical assistance program to make emerging economies more robust to AI labor market disruption, or efforts to build clinical trial and regulatory infrastructure in low- and middle-income countries (LMICs) before AI-enabled treatments outpace the systems meant to deliver them.
We are open to funding a wide range of work, including writing, policy analysis, and research, but we’re especially excited by the prospect of seeding new organizations and supporting direct implementation.
We expect to allocate $10-30 million through this RFP. We plan to accept applications until August 21st, and will review applications on a rolling basis, though we expect to take longer to review larger grants. You can apply to this RFP here.
Background
AI is progressing rapidly and could have profound implications for the health and wellbeing of people around the world. While global health philanthropy has historically assumed relative continuity with the past and not been closely tied to advances in AI, we would like to engage seriously with the implications of transformative AI for global health and economic prosperity.
This RFP supports rigorous and cost-effective research, policy development, field-building, and implementation focused on improving health and economic outcomes for individuals in measurable ways that are responsive to both the challenges and opportunities that a world transformed by AI might bring. Please see the example projects below. We’re interested in outcomes that are not catastrophic or human extinction-level but still discontinuous — distinct from the focus of our Navigating Transformative AI Fund, which addresses potential catastrophic risks.
This RFP takes seriously the possibility that AI reshapes which global health problems are tractable and where the highest-impact opportunities lie. That said, we think the questions core to global health work remain the same: who is helped, by how much, at what cost, and compared to what alternative?
A few examples of what a transformative AI future might look like in terms of R&D and health:
- A “compressed 21st century” in biomedicine. AI could compress many decades of biological progress into a few years, plausibly delivering cures or effective treatments for a wide range of diseases. The binding constraint would then shift from scientific discovery to regulatory approval and distribution, particularly to ensure new treatments reach the global poor.
- A shift in where the bottlenecks to R&D sit. AI is likely to accelerate stages of the biomedical pipeline unequally — dramatically speeding up literature synthesis, target identification, molecular design, and computational biology while leaving wet-lab work, clinical trials, regulatory review, and manufacturing scale-up much less affected. Some investments that are important in this world have long lead times to prepare, like clinical trial infrastructure in LMICs, regulatory approval capacity and streamlining, and data infrastructure that represents underrepresented populations.
And some examples in economic terms:
- AI drives explosive economic growth. If AI automates R&D and a substantial share of labor, rapidly eliminating global poverty could become technically possible. But the distribution of those gains would not be guaranteed. The bottleneck on improving lives may then shift to barriers to distributing funding, general political support for higher levels of aid, and specific proposals for how to ladder up assistance as gains from AI materialize.
- AI erases a rung of the canonical development ladder without an obvious replacement, increasing cross-country inequality. AI could widen disparities between LMICs and advanced economies by accelerating reshoring, reducing foreign direct investment in LMICs, and closing off service exports as remote work (e.g. call centers, software outsourcing) is automated. In this world, the demographic dividend in places like South Asia and sub-Saharan Africa risks becoming a liability as labor tax bases shrink, and the financial foundations of LMIC health and welfare systems weaken. Preparing for this world might look like building institutional absorptive capacity, supporting policy work on transition management and alternative growth pathways, or strengthening LMIC tax systems in anticipation of shrinking labor tax revenue.
These are just a few examples of potential outcomes, and the scenarios could overlap and blend together. We’re interested in work responsive to any of them, to combinations, or to futures we haven’t outlined.
We think there’s significant uncertainty about which world we actually end up in, and what the second- and third-order consequences look like. We don’t expect to resolve that uncertainty (or for applicants to do so either!). But we believe that seeding a field of work that engages with this range of possibilities is valuable, and we expect to make differentiated bets responsive to different outcomes.
Through this lens, we are interested in work that both minimizes the negative impacts that AI might have on health and wellbeing as well as work that maximizes its potential benefits.
We don’t have a firm thesis about what the strongest proposals will look like, though we outline some ideas below. We expect the best applicants will teach us something we hadn’t considered, or surprise us with new ideas.
What we will fund
We are excited about funding work along two complementary axes: identifying the preparatory steps global health funders and actors should take (research, policy analysis, empirical mapping), and executing on preparatory steps that are likely high-value (infrastructure, relationships, field-building, piloting). We think LMICs will likely be especially vulnerable to the downsides and especially poorly positioned to capture the upsides, and that work focused on impacting these countries will likely be more neglected. That said, we are fully open to work focused on high-income countries, provided it responds to the broader orientation above.
A few points we’d like to note upfront:
- We’re particularly interested in seeding new organizations or bringing new people into thinking about this area.
- Because there is a huge uncertainty on how transformative AI scenarios could develop, we’re more interested in supporting people to work on key aspects of the problem than funding pre-specified pieces of academic research papers. Most research we fund will include a public communication and/or policy engagement component; we are unlikely to fund projects without these features.
- A key distinction we’d like to make is that this RFP is focused on the larger implications for health and wellbeing of a world drastically transformed by AI. It is not geared toward funding “AI for good” interventions — projects where AI is directly utilized as a tool to improve health, education, agriculture, etc. That said, we plan to continue supporting uses of “AI for good” tools through our regular grantmaking, and may run a follow-on RFP focused on this in the future.
We expect to allocate $10-30 million through this RFP. Here are a few indicative categories:
- Exploratory grants (less than $100k): Research and writing projects, scoping studies, prototypes of potential future work, and short-form policy work.
- Standard grants ($100k to $1m): New hires at an existing organization, larger research programs, or launching a small (1-3 person) organization.
- Large grants ($1m to $10m): Larger new organizations, institutional capacity-building, substantive policy engagement, or more complicated preparatory infrastructure.
- We’re open to larger proposals, but they would likely require a longer review process outside this RFP.
Here are example ideas of specific grants for illustrative purposes, though we are open to many others and we are not committed to funding applicants who apply with these ideas:
Exploratory grants (<$100k)
- Biopharma pipeline bottleneck mapping. A short report from a biopharma R&D practitioner mapping which stages of drug development are most accelerable by AI and which become binding constraints under aggressive AI assumptions. This might look like a transparent quantitative model and a public essay.
- Research on how health and wellbeing cost-effectiveness analyses and grantmaking should change in transformative-AI scenarios. Examples might include significantly different weights on health relative to income, or deprioritizing areas that might be sped up by AI while concentrating on those more likely to be bottlenecks or enablers for it.
- LMIC AI-exposure occupational mapping. A map laying out which occupations are vulnerable to AI for LMICs, adapting existing U.S.-centric methodologies to LMIC labor force composition and informal sector realities, e.g. performed by an LMIC-based think tank with strong policy relationships.
- Research to identify key data public goods. Identifying what public goods, particularly in the form of data, AI would make significantly more valuable and assigning an importance rating and cost of development to each (to be shared publicly).
Standard grants ($100k-$1m)
- LMIC tax base resilience research. A 12-18 month research program examining tax base design under shrinking labor tax revenues — VAT, digital services taxation, and capital income taxation under weak administrative capacity. Country case studies in 3-4 high-stakes contexts. This might look like a flagship public paper and country-specific working papers intended to inform both donor strategy and LMIC government policy.
- Quarterly AI-labor-impact tracker for exposed LMIC sectors. Pilot for panel-based quarterly tracking of AI’s effects on income, hours, and employment in business process outsourcing, software outsourcing, transcription, and content moderation, housed at a think tank or major global institution.
- LMIC AI industrial strategy research and policy engagement. A 12-18 month research and policy program developing concrete alternative growth strategies for countries whose traditional development paths — export manufacturing, services outsourcing — are foreclosed or weakened by AI. Goes beyond diagnosis to propose which sectors specific countries should bet on, what policy instruments to use, and how to sequence reforms. This might look like a flagship report, country-specific policy notes, and/or convenings.
- Policy fellowship on transformative AI and global health. Funding for an experienced policy fellow to be embedded at a host institution (e.g. a think tank, multilateral, or LMIC government body) to work at the intersection of transformative AI and global health and wellbeing. Possible focus areas include accelerated clinical trial frameworks, LMIC government engagement on AI policy, regulatory harmonization for AI-discovered therapeutics, or aid architecture under high-budget scenarios.
- Ph.D. student bootcamp on AI and LMICs. Funding for a group of Ph.D. students to gather for 1-2 weeks and learn about current research on AI and economics/public health/political science as it relates to LMICs, and discuss AI and global health and development research topics.
- Building a data access layer. Setting up proposals for data-sharing frameworks, consent, and privacy protections to enable cheaper and faster impact evaluation of GHW-style interventions, including RCTs and monitoring. Identifying and digitizing useful pockets of data.
Major grants (>$1m)
- Seed funding for a new policy organization on accelerated biomedical R&D and access. A small organization focused on what major global health funders and multilaterals should be doing now to prepare for a “compressed 21st century” in biomedical R&D, including IP and access frameworks for AI-discovered therapeutics, governance of cross-border training data, and pull mechanism design.
- Design study for a public-private partnership on AI benefits to the global poor. An effort by a small team to do the governance, legal, and political-economy pre-launch design work for a global multilateral institution focused on ensuring AI benefits reach the global poor.
- Organization to research and implement new, faster methods of evidence generation. AI might lead to a flood of new ideas and potential interventions, but RCTs are expensive and time-consuming. This project might identify how AI-assisted (and fully consented) RCTs could be dramatically cheaper and quicker (e.g. by analyzing large datasets like mobile money, satellite, telecoms, or government admin data) and then pilot this in 1-2 contexts.
- Organization focused on mobile connectivity and affordability in LMICs. This might include levers like competition policy and regulatory-capacity support, or policy advocacy work.
- Seed funding for a digital public infrastructure audit and gap-filling fund. Commission assessments of digital public infrastructure readiness (digital ID, payment rails, data infrastructure) in high-exposure LMICs, paired with seed funding for the highest-priority gaps, meant to strengthen both service delivery resilience and revenue collection under AI disruption.
- Regulatory capacity-building for AI-era financial and labor systems. Grants to strengthen LMIC regulators’ ability to oversee AI-driven financial services, gig platforms, and automated hiring.
- Technical assistance program for adaptive social protection design in AI-exposed economies. Fund a small team embedded within an LMIC government to help redesign cash transfer targeting, eligibility, and financing mechanisms that are robust to labor market disruption, translating tax base resilience research into actionable government support.
How we assess grants
Since these grants are intended to address a wide spectrum of future scenarios, we don’t think that there’s a straightforward template for a successful proposal. However, we often use the Importance, Neglectedness, and Tractability framework, which in this RFP might cover:
- Importance: Would success meaningfully sharpen our understanding of, or capacity to respond to, transformative AI impact? Does this work address a major implication of transformative AI?
- Neglectedness: Is this work that would not happen, or would happen slower or worse, without our funding? LMIC-focused work is relatively neglected, for example.
- Tractability: Is the proposed approach likely to produce useful outputs on the stated timeline? Does the team submitting the proposal have relevant past work, skills, and networks?
Because we assess grants through their expected impact on health and economic prosperity, if you apply with a research or writing proposal, we encourage you to consider and clarify how you expect that work to lead to direct impact (e.g. it is very likely to inform a particular stakeholder in making a particular high-stakes decision). We care quite a bit about this, and are very unlikely to fund research without a clear path to impact.
How to apply
We plan to accept applications until August 21st. We plan to review applications on a rolling basis, though we expect to take longer to review larger grants. You can apply through this form. The application will ask whether your requested grant size falls into the exploratory, standard, or major categories, and the process from there will differ:
- Exploratory grants (<$100k): We will get back to you with a decision.
- Standard grants ($100k-$1m): We will reach out with next steps, which may include follow-up questions or a direct decision.
- Major grants (>$1m): This is the first stage of the process, which is an expression of interest. We will review these applications monthly and invite select applicants to submit a full proposal.
By default, grants will last up to 2 years. Their renewability will depend on the nature of the project (e.g. funding to found an organization will typically be renewable, whereas funding to produce a discrete piece of work will not be).
Eligibility: Open globally to individuals, academic institutions, think tanks, nonprofits, and (where structured appropriately) for-profit entities. We welcome applicants based in or with deep working knowledge of LMICs, and those who have not previously worked on AI futures. Proposals to spin up new organizations are in scope.
Confidentiality: If we’re unable to provide support but think your project is promising, we may share your application materials with other funders who may be interested. Let us know if you’d prefer we not do this.