High-Impact Growth – "What Makes a Dollar Matter? Lessons from Coefficient Giving"
Podcast Host: Dimagi (Amie Vaccaro & Jonathan Jackson)
Guest: Dina Moussa, Lead Researcher at Coefficient Giving (formerly Open Philanthropy)
Date: December 10, 2025
Episode Overview
This episode explores what it means to make every philanthropic dollar count, guided by the evidence-based, impact-maximization approach of Coefficient Giving (formerly Open Philanthropy). Host Amie Vaccaro and Dimagi CEO Jonathan Jackson sit down with Dina Moussa to unpack the frameworks and practices behind Coefficient’s cause prioritization, cost-effectiveness research, and experimental grantmaking—plus the evolving role of AI as a 'force multiplier' for global health, equity, and development impact.
Guest Introduction and Motivation
[02:12–05:50] Dina Moussa’s Career Path and Philosophical Approach
- Dina shares her academic background (philosophy and economics at Yale), her attraction to the “radically different trajectories” of countries and communities, and what drew her from consulting into global health philanthropy.
- She emphasizes “the power law applies to doing good; some opportunities for impacting people’s lives are just thousands of times more cost-effective than others.” (05:32)
“Somewhat small or seemingly unimportant differences in how systems or organizations are set up can lead in the long term to these really radical different outcomes.”
— Dina Moussa [03:25]
What Sets Coefficient Giving Apart
[05:50–07:14] Neutral, Evidence-based Impact Maximization
- Coefficient Giving takes a neutral, analytic approach; it's open to “any cause area” based on empirical impact per dollar, distinct from most foundations’ advocacy-driven or association-based giving.
- Their approach is modeled along the lines of GiveWell but expands to broader policy and R&D domains.
“In your field, in your sector, are funders agreeing with the premise that like, their job is to maximize the impact per dollar spent, and that is something that sets apart...the impact maximization foundations.”
— Jonathan Jackson [07:14]
Cause Areas and Grantmaking Strategy
[08:20–09:36] The Four Pillars of Funding
- Four main domains:
- Direct Health Interventions: e.g., mosquito nets, vaccines (via GiveWell)
- Global Health Science and R&D: neglected diseases and market failures
- Policy Work: macroeconomic growth, environmental health, more efficient aid
- Abundance & Growth in High-Income Countries: e.g., housing reform
[09:52–12:39] Cause Prioritization & Cost-effectiveness in Practice
- Uses “social return on investment” (SROI): models how much health or income benefit is delivered per dollar spent.
- Grants are made if they cross a minimum bar for benefit-to-cost ratio, attempting budget parity so each dollar does the most good across programs.
“We try to make it such that the social return on the last dollar spent in every program is the same. And so that there’s no way you could move money from one program to another in a way that would do more good.”
— Dina Moussa [12:20]
[13:16–17:39] Deciding What to Fund: Importance, Neglectedness, Tractability (INT Framework)
- “Importance” = scale/intensity of the problem.
- “Neglectedness” = how much funding/attention it already receives.
- “Tractability” = is it actually solvable by philanthropy?
- Example: Investing in economic growth in LMICs: huge potential, but hard to get right (need strong conviction in “importance”).
- Example of pausing on new programs: Data for LMICs—tractable consumer/use case not clearly identified; sometimes evidence base isn’t strong enough for actionable grants.
“There are also cases...where at one point we investigate an area and we just don’t feel comfortable enough, but we come back to it a few years later...so it’s not necessarily one stagnant decision, but more of an evolving view...”
— Dina Moussa [15:52]
"Learning As Fast As Possible" as a Funder
[18:11–20:42] Agile, Iterative Grantmaking & Research
- Maintains “a long list” of potential cause areas and uses rapid “cause sprints” (2 days) or shallow investigations (2-3 weeks) to prioritize.
- Early grants are made to maximize learning (“value of information”)—not just delivering outcomes, but revealing how to do better in future allocations.
“Those first few grants are often chosen with learning in mind...we might try to make smaller grants across several organizations with different theories of change...and use that to come to a consensus on what the right approach is.”
— Dina Moussa [20:05]
Deep Dive: AI’s Role in Global Health & Development
The Radiology Parable
[21:54–27:39] Why AI Hasn’t Replaced Radiologists
- Despite predictions, radiologists’ jobs are not (yet) automated away; in fact, demand and wages have increased.
- Three key reasons:
- Technical Limitations: Models don’t generalize across all settings (overfitting on hospital-specific details).
- Regulation & Liability: “Who’s responsible” when an AI makes a mistake isn’t sorted.
- Economic Forces: Jevons Paradox—greater efficiency led to more usage, not less demand for radiologists.
“As you can imagine, this generally sped up their work...But instead of causing a reduction in radiologists, it actually caused an increase in scans...”
— Dina Moussa [26:11]
Lessons for AI Integration in Global Health
[28:17–32:34] Context-Specific, Equity-Centric Adoption
- Broad, benchmark-based predictions of AI job disruption are too simplistic; “it’s hard to predict the impact...unless you deeply understand those fields” [28:34].
- In LMICs or under-resourced settings, “the alternative to a medical AI model might not be going in to see your primary care physician—it’s maybe not going to the doctor at all.”
“In these contexts, I think the question of what AI can do and what gap it is filling is very different.”
— Dina Moussa [31:23]
- Raises the ethical tension: Is “good enough” (80% of a human’s performance) worth deploying when the baseline is nothing at all?
How Does Coefficient Giving Approach AI as a Cause?
[32:42–34:40] Sectoral & Horizon Scanning Approach
- Rather than forcing AI as a solution, Coefficient Giving first asks, “what bottlenecks does AI uniquely unlock?” and then looks across verticals to spot where AI could be catalytic.
- Rapid shifts in the technology require continuous reassessment.
Equity, Language, and Benchmarks in AI
[35:14–39:09] Improving AI for Low-Resource Languages
- Current models perform worse on low-resource languages, but fine-tuning is relatively cheap and high-impact—especially for health outcomes.
- Philanthropy can step in where “market incentives” under-serve, such as funding data collection or improving benchmarks relevant to LMICs.
“It’s an example of technological path dependency...new technology is built with its local context in mind...it’s not quite right for [lower-income] countries.”
— Dina Moussa [35:46]
The Role of Philanthropy vs. Industry vs. Government
[40:50–42:16] Who Does What?
- Rapid field evolution means roles for philanthropy, government, and business are fluid; more cross-sector coordination is vital to ensure impact.
- Philanthropy should look for “relatively low cost but high leverage” interventions—like setting benchmarks and building data for neglected contexts.
Five-to-Ten Year Outlook: Optimism & Caution for AI in Global Health
[42:16–45:53] What's Possible—and Where to Be Careful
- Most exciting hope: AI as a “force multiplier on human capital”—helping individuals and bureaucracies work more efficiently, amplifying existing roles rather than replacing them.
- Cautions: Unintended safety risks, rushing to deploy untested tools, or getting too excited about benchmarks before clinical/real-world efficacy is proven.
“I worry a little bit about getting really excited about...use cases, which maybe makes us excited too quickly about work that has a lot of...potential ramifications...”
— Dina Moussa [44:47]
Final Takeaways: How Should Global Health Leaders Think about AI?
[46:24–47:15] Don't Presume the Answer is Always AI
- Leaders should start by asking, “what is truly special about AI, and where is it most likely to be helpful?”—not force it into every problem (“hammer in search of a nail”).
- Demos and new features will proliferate, but “ground all this work in how you're going to measure it, how you're going to improve it.” — Jonathan Jackson [47:37]
- Stay anchored in proven, measurable impact even as the technology landscape shifts rapidly.
Notable Quotes & Moments
- “Some opportunities for impacting people's lives are just thousands of times more cost-effective than others, more impactful than others.”
— Dina Moussa [05:32] - “Our job is to maximize the impact per dollar spent.”
— Jonathan Jackson [07:14] - “We call this a social return on investment...modeled after the private sector term...Across our cause areas, we have a sort of minimum bar for the benefit to cost ratio a grant should have...”
— Dina Moussa [11:48] - “So that there's no way you could move money from one program to another in a way that would do more good.”
— Dina Moussa [12:25] - “AI is not a magic fix but rather a force multiplier. Its value depends on how deeply we understand the systems it touches.”
— Amie Vaccaro [51:11, show closing] - “Don’t presume the answer is AI—start by asking what is truly special about AI and where it will be most and least helpful.”
— Dina Moussa [46:27] - “There’s going to be an amazing explosion of AI demos...it’s really important to ground all this work in how you’re going to measure it...not getting caught up in the next shiny thing.”
— Jonathan Jackson [47:23–47:44]
Key Timestamps
- 02:12 – Dina’s journey into global health & philanthropy
- 05:50 – The “impact maximization” philosophy & approach
- 08:20 – Coefficient Giving’s four funding focus areas
- 09:52 – How cause prioritization and SROI works
- 13:16 – The INT framework (importance, neglectedness, tractability)
- 18:32 – “Learning as fast as possible”: cause sprints and exploratory grants
- 21:54 – AI & radiology: why disruption predictions were wrong
- 28:17 – Lessons for integrating AI in global health
- 31:23 – AI in LMICs: Different standards, different expectations
- 32:42 – How Coefficient Giving evaluates AI as an intervention
- 35:14 – Low-resource languages, equity, and AI training data
- 40:50 – Philanthropy’s role vis-à-vis industry and government in AI
- 42:16 – AI in global health: promise, applications, and risks
- 46:24 – Advice to global health leaders re: AI
For Further Follow-Up
- Dina Moussa’s newsletter: "Under Development"
- Find more episodes: Dimagi Podcasts
Summary prepared to capture the episode’s content, tone, and actionable insights for listeners seeking to advance evidence-driven, high-impact philanthropy in the age of AI.