Project Evident: AI and Bias

The Project

Project Evident helps nonprofits and other social-sector organizations use data and evidence to do more good. As more of those organizations start using AI, Project Evident wanted a short, plain-language series that shows what responsible AI use looks like in a real nonprofit setting. RaffertyWeiss Media produced the three-part animated series. AI and Bias is the first episode, followed by AI and Privacy and AI Agents.

Our Approach

Every episode takes place at Housing for You, a fictional nonprofit in Springfield, USA. Staff members work alongside Ada, an AI sorting tool, and Adam, a generative AI writing tool. Showing AI as tools with clear limits, not as people, keeps the lessons concrete. In this episode, Ada ranks a rental assistance waitlist and leaves out the families who need help most. The team traces the problem to three sources of bias:

  • Incomplete information. Shelter walk-in applications were never entered, and an AI tool "cannot see information you never gave it."
  • Biased rules. Ranking on credit score and verifiable income shut out clients who are paid in cash.
  • Learning from people. Adam, trained on the internet, writes a denial letter full of assumptions no one asked for.

The episode ends with the fix: complete the data, set rules that match the mission and have a person review every AI output before it is used. Its closing line sums up the series: "Concerns about bias doesn't make AI useless. It makes you necessary."

What We Delivered

  • A 5½-minute animated explainer, the first in a three-part series on responsible AI
  • Character design and storyboards for a recurring nonprofit cast and two AI tools
  • Professional voice casting and direction for a six-person voice cast
  • Final animation, sound design and mix

Related Services