AdobeStock_320460723.jpeg

The Future of Grant-Making: How AI is Reshaping Proposal Evaluation

The philanthropic sector faces a mounting operational challenge: how to efficiently evaluate an increasing number of grant proposals while maintaining rigorous standards. The traditional solution of adding more reviewers or extending review periods isn't sustainable, especially for organizations committed to rapid response funding or regular grant cycles and operating with limited resources and staff.
Published on by Pooja Wagh and Hala Hanna

Never miss an update from our AI for Good series: sign up for the AI for Good newsletter.

A research collaboration between MIT Solve, Harvard Business School, the University of Washington, and ESSEC Business School is currently exploring this challenge, using MIT Solve’s application review process as a beta. The initial findings offer valuable insights for foundations considering AI integration in their evaluation workflows. 

Testing AI in Grant Review: What Works?

To evaluate application screening, the research team developed an AI system based on GPT-4 and tested it with different types of reviewers, ranging from very experienced to not experienced at all. The findings reveal both promising applications and important limitations of AI in grant review.

  1. Objective Criteria Assessment: AI tools proved most reliable for screening basic eligibility requirements and alignment with funding priorities. For instance, it could accurately identify whether proposals included required elements or matched stated geographic or programmatic focus areas.

  2. Reviewer Support: Less experienced reviewers made more consistent decisions when supported by AI insights, suggesting the potential for expanded review capacity that maintains quality standards. However, seasoned reviewers used AI more selectively, often as a supplementary tool rather than a primary guide.

  3. Standardization Benefits: While AI didn't significantly speed up individual application reviews, it helped standardize the screening process and improved consistency across different reviewers.

A Practical Path Forward

The initial research suggests a promising two-stage approach that many foundations could adopt:

  1. Initial screening using AI to evaluate objective criteria

  2. Human review focused on nuanced aspects like potential impact, innovation, and organizational capacity

This hybrid model has the potential to allow foundations to:

  • Handle increased application volumes more efficiently

  • Maintain high standards of due diligence

  • Deploy reviewers' expertise more strategically

  • Expand reviewer pools without compromising quality

Looking Ahead: Implications for Philanthropy

As the sector grapples with growing demands for both efficiency and thoroughness in grant-making, thoughtful AI integration offers promise.  Early research suggests that AI can enhance—rather than replace—human judgment in the review process.

However, important questions remain about best practices for implementation. How can foundations ensure AI tools don't perpetuate existing biases? What training do review teams need to effectively collaborate with AI systems? How might these tools affect applicant experience and outcomes?

As more foundations experiment with AI integration, sharing lessons learned will be crucial. The philanthropic sector has an opportunity to shape how this technology serves its unique needs and values, ensuring that efficiency gains don't come at the expense of careful, mission-aligned grant-making.

The early evidence from this research study suggests that AI when thoughtfully implemented, can help foundations process more applications while maintaining the careful consideration that impactful philanthropy requires. For a sector often caught between growing demands and limited resources, this presents an intriguing path forward.

Share this article: