Board Case Study · Nonprofit (Foundation)
To remove bias, the foundation let an AI rank grant applicants. It funded higher-scoring groups, and defunded the small, community-led organizations at the heart of the mission.
The situation
A large philanthropic foundation, seeking to reduce human bias and process more applications, deployed an AI system to score and rank grant applicants on predicted impact. It funded organizations with strong measurable outcomes, and systematically downranked small, community-led groups that lacked the data and polish to score well.
Those grassroots organizations were central to the foundation's stated mission and its commitment to equity. The model optimized a proxy for impact it could measure, not the mission it was meant to serve. Grantees and the field are noticing the shift.
The board must weigh efficiency and consistency against mission drift, the equity implications of algorithmic gatekeeping in philanthropy, and its fiduciary and charitable duty to fund toward purpose, not just metrics.
“We built it to be fairer. It funded the polished and starved the grassroots, exactly the groups we exist for. Did we optimize away our mission?”
Board Chair
The decision on the table
The full case, scenario architecture, board materials, and facilitation notes, is shared with boards and partners on request.