Impact and trust: Key takeaways on nonprofit AI practices

Levels of AI adoption among nonprofits is wildly uneven. While some are making AI solutions an integral part of their mission, many are using it ad hoc for routine tasks with no organizational AI policy. A report from the Blackbaud Institute asks: How can nonprofits invest in AI practices most effectively and responsibly to amplify their impact and strengthen donor trust?

Here are a few takeaways from Bridging the AI Effectiveness Gap​; New Research on What Drives AI Impact and Trust in the Social Sector, based on surveys of nonprofit staff and donors.

Using the Responsible AI Institute’s five stages of Responsible AI Maturity, the report categorized just one in 10 of participating nonprofits as “AI-adaptive.” That means those nonprofits’ AI practices have moved beyond experimentation to systemic application of responsible practices and processes aligned with strategy and mission. By contrast, nearly four in 10 were merely “aware” nonprofits—those using AI but without any formal structure, coordination, or documentation.

1. AI-adaptive nonprofits see returns on investment in AI

Nonprofits that fully integrated and aligned AI with strategy and mission were significantly more likely than others to report positive results in fundraising, donor acquisition, and retention. AI-adaptive nonprofits were more likely than other nonprofits to see increased overall revenue (75%), improved fundraising revenue (64%), and strengthened donor retention (48%).

2. Impact depends on how nonprofits reinvest AI’s time savings

Nonprofits at all stages of AI adoption said AI saves staff time—$500/employee/week on average. But AI-adaptive organizations saved more, $621/employee/week and, more importantly, reinvested that time and staff capacity to drive revenue and impact.

AI-adaptive nonprofits were significantly more likely than others to say AI freed up time to focus on mission-critical work, deliver services more effectively, and raise more money. At nonprofits without organization-wide AI practices, time savings may help individuals but don’t translate into overall impact. AI-adaptive organizations are better positioned to reinvest capacity into work that drives results: improving data quality, strengthening donor trust through transparency, and building out nonprofit AI practices.

3. Four critical gaps divide nonprofit AI practices

The study identifies four gaps between AI-adaptive nonprofits that are realizing significant benefits and organizations that are not yet AI-adaptive:

The effectiveness gap. While 85% of respondents said they used AI at work, only 33% said their organization was using it very effectively. Individual experimentation does not always lead to effective organization-level adoption. AI use varied by generations or roles, with millennials and those in leadership positions using AI more frequently and experiencing more benefits.

The infrastructure gap. Without organization-level AI adoption, most staff were using AI in fragmented ways—with limited impact. Many nonprofits have not yet built infrastructure for scaled, governed, safe AI practices. Barriers included lack of staff knowledge, training, and skills; limited resources; and concerns about data accuracy and security. Only half had paid or enterprise versions of AI tools; nearly a quarter used free versions, exposing themselves to risk.

The data-readiness gap. Data readiness is the foundation of AI readiness. Staff at AI-adaptive nonprofits were about twice as likely as respondents overall to say they were very confident about their data’s accuracy, validity, timeliness, completeness, consistency, and uniqueness. AI-adaptive organizations were also more likely to have dedicated data management staff, and 81% had experimented with or deployed AI to improve data quality.

The transparency gap. Nonprofit practices consistently fell below donors’ expectations for transparency and data security. More than three-quarters of donors said it’s important to disclose when and how AI is used, but only about a quarter of nonprofit staff said they do so. There were similar gaps between donor priorities and nonprofit actions around protecting sensitive or personal data (68% vs. 36%), requiring human review of AI-generated outputs (61% vs. 42%), and assessing AI systems for risks in bias, accuracy, or data security (60% vs. 28%).

4. Closing AI gaps will help advance the sector

The report’s authors note these gaps in efficiency, infrastructure, data readiness, and transparency underscore what they call “a critical reality”: Nonprofit AI practices that are not governed, transparent, and aligned to donor expectations risk eroding trust—at a time when the sector needs to strengthen that trust.

“The path forward is not to adopt more AI for its own sake, but to close the gaps that prevent adoption from translating into organization-level results” by taking targeted steps to close each gap, the authors conclude.

This article is part of a regular feature where Candid insights shares key takeaways from a new research report to encourage a more data-driven approach to the sector’s work. Please email insights@candid.org to recommend a report for an upcoming feature. 

Photo credit: JohnnyGreig/Getty Images