Call for Concepts: AI for Human Agency

Overview

 The Agency Fund is a nonprofit partnership investing in ideas and organizations that help people navigate toward a better future. We envision a coach, a mentor, a counselor for all. We maintain an open funding window to identify research and innovation concepts within the general scope of our approach. We also partner with funders to issue more targeted calls for proposals (such as this one) in specific topic areas.  

Contextual Background. Much of the philanthropic conversation around AI has been focusing on its risks: its potential to manipulate or mislead people, render their capacities obsolete, and more generally diminish human agency.  However, AI also has the potential to augment human agency. The creation of The Agency Fund was motivated by a rich body of encouraging evidence on the impacts of social workers, role models, mentors, and similar “support personas” that help people navigate their own lives. The unifying idea behind these approaches is to empower people – augmenting their decision-making, rather than over-riding it.

We are interested in projects that capitalize on recent leaps in generative AI to deliver personalized, interactive support, especially in lower-income settings. We are especially interested in generating evidence that helps determine the cost-effectiveness of such approaches.

What and How We Plan to Fund

We seek to fund 6-12 month projects that leverage AI to help people navigate their lives, ultimately enabling them to improve their economic, health, or educational outcomes. We are particularly interested in projects that meet the following criteria:

• Performance Data & Evidence Generating. The project comes with a credible (likely experimental) approach to assess performance on user engagement, proximate outcomes (e.g., beliefs / knowledge), as well as downstream social & development outcomes (e.g., educational or economic metrics).  
• Agency Enhancing. The innovation embraces people’s consciousness and helps them hone their beliefs and mental models without over-riding their judgment. We are interested in academic innovation (e.g. in the psychological, behavioral, neuroscience, decision, economic, or computer sciences), but also in specific use cases (say, helping US domestic violence survivors escape abusive relationships; women in Nigeria navigate safe childbirth; students in Peru plan their educational journeys; Indian citizens make sense of their entitlements; etc.)  
• Safe and Responsible. The innovation identifies and commits to appropriate guardrails to manage the risk of misinformation, privacy violations, and other harm. Also, the research complies with ethical frameworks on human subject research.
• Scalable.  There is a credible path to scale (1m+ users) via policy mechanisms, networks, or a business model.      

We expect that the most competitive projects will be contextually adapted and tailored to specific populations and decision challenges. This could involve the use of dedicated training data; tailored user experiences; or other adaptations.  While our partnerships and collaborations can assist promising projects in bridging certain scientific and technological gaps, a thorough understanding of the local context is indispensable. Applicants will need to demonstrate intimate familiarity with the populations and decision challenges they aim to support.         

We only accept one application per primary contact in any given quarter.

We expect to allocate $2-4M to this funding round. We expect that most projects will be pilots in the $200k to $400k range.