A Foundation Pilot Program Simplifies Grant Reporting With AI

Staff at the Houston Endowment were acutely aware of how much grantees disliked the grant-reporting process. The topic comes up often — at conferences, in reports, and pretty much everywhere in the nonprofit sector.

With that in mind, last year the Houston Endowment — a major funder of Houston groups focused on issues like public education, the arts, and creating public spaces — decided to make a change. It launched a pilot to use Zoom AI tools to record and transcribe virtual meetings between program officers and grantees from its civic engagement program.

“We knew there was a difference in the quality of learning and information that you can get from an oral report versus a written report,” said Robiel Abraha, a learning and evaluations officer who helped create the pilot. “When you’re allowed to ask follow-up questions and give grantees the opportunities to tell stories, you can add a lot of nuance and context that might not be in a written report.”

Previously, the Houston Endowment required grantees to write responses to a list of questions through an online portal, which could require several hours of work, according to endowment grantees. The oral interviews, on the other hand, usually lasted about 45 to 90 minutes and could include multiple people from the grantee organization, reducing the amount of time spent gathering relevant information from staff members. Grantees gave permission to be recorded and were provided discussion questions in advance.

Those conversations have led to a better understanding of how nonprofits are using grant dollars, said Miguel Rivera, a program officer for the endowment’s civic engagement program.

“There was a greater ability for me and for our team to better identify what was going on with our partners and also share what trends we were seeing more broadly across the field — something that was harder to do when we were just getting written reports,” he said.

The Houston Endowment’s pilot could help set a precedent for other foundations and grant-maker groups, said Elisha Smith Arrillaga, vice president for research at the Center for Effective Philanthropy, which produces data on the sector.

“One of the powerful things about the Houston Endowment example is that they listened when grantees said, ‘Hey, this process isn’t working for us.’ And they are testing this strategy to see if it works better,” she said.

The AI Pickup Problem

Now that the AI reporting pilot has concluded, the Houston Endowment is expanding its oral reporting to other programs, such as its arts and culture grant-making portfolio. The organization recently presented findings at a recent Grantmakers for Effective Organizations national funders conference, where attendees were eager to learn more about the initiative, Abraha said, and the endowment was eager to see fellow grant makers experiment with AI tools that help grantee reporting.

But that could be a while off. This week, the Center for Effective Philanthropy, a group that researches giving trends, published a report that found while some AI use among foundations and nonprofits was common — such as for drafting emails and documents — staff at these organizations didn’t fully understand the technology, and formalizing AI use within daily operations had been slow. Many groups had concerns about the accuracy of the technology, security risks, and the lack of AI expertise among their staff, according to the report, which was based on surveys of 215 foundation leaders and 451 nonprofit leaders.

The majority of foundation leaders also said their staff didn’t understand what grantees’ AI needs might be. Few funders were providing any form of support, monetary or nonmonetary, to help nonprofits use AI in their operations. And just over a third of grant makers said they had received few, if any, requests for AI support.

There were broad concerns about the potential for AI use to go awry, CEP found. More than 80 percent of foundation leaders said they worried about possible data and privacy risks, and 73 percent of nonprofit leaders said they were concerned that AI technologies could produce misinformation or inaccurate results. And some expressed concerns that biases and discrimination baked into AI algorithms could negatively impact underserved communities.

The sector needs to “look at how it affects the communities that do not represent those creating the AI,” one nonprofit leader told CEP. Names of respondents were not disclosed in the report.

Yet if philanthropy leaders want to ensure the technology is developing in ways that benefit all communities, then the CEP report shows the sector is missing an opportunity to do that because they are not engaging with the technology, Smith Arrillaga said.

Talking Through AI

Trust between grantees and program officers has been critical to the success of the Houston Endowment’s AI reporting pilot, according to grantees. Frequent check-ins were already part of the relationship between the endowment and Mi Familia en Acción, a group that helps Latinos register to vote in eight states, according to Dayana Iza, Mi Familia’s state director for Texas. When the endowment introduced the idea of switching to oral reports, Iza said she wasn’t concerned that they would misuse the information. Rather, her main concern was that AI might not be able to capture “the essence” of the work her group was doing.

“I wanted to make sure that it didn’t lose the essence of being real, being human, because I can usually tell when something is written by AI,” she said.

Iza added she has been relieved to learn that Endowment staff review transcripts of the oral reports and correct any misspellings or other errors.

Katie Campbell Shumway said she also was delighted when she learned she could complete grant reporting for the League of Women Voters of Houston with a virtual call rather than spending hours trying to pull together information for a written report, which may not fully convey the work that was done with the grant money.

“Any chance you get to have a conversation with your grants officer is like a golden opportunity,” she said. “So to even have one officer dedicated to talking about the reporting of our work and how things came together throughout the process was really helpful.”

The Houston Endowment also allows for grant dollars from the Collaboration Fund that it set up last November to help cover organizations’ costs for technology, including AI, foundation spokesperson Stephanie Getman said. Last year, the endowment distributed nearly $5 million in grants through the Collaboration Fund. This year, it delivered $15 million.

Some of the grantees are already exploring how they also might incorporate AI into their own operations. The league has used some its grant to pay for a tech expert to review best practices for using AI, according to Campbell Shumway, who is the group’s executive director. Mi Familia en Acción has installed chatbots on its website to help provide voting information to communities, according to Iza.

Iza said she has learned a lot more about potential AI uses from the oral reporting pilot. Her takeaway: “It does save us a lot of time. And I hope other foundations explore the idea of what this could look like for them.”