More than two-thirds of nonprofits have already tried using artificial intelligence in their work, despite low comfort levels with the technology and concerns about data protection and bias, a new survey reveals.
The report, released Tuesday by GivingTuesday’s Generosity A.I. Working Group, paints a nuanced picture of a sector grappling with the promises and perils of A.I. adoption. The study surveyed 930 nonprofits in the hashtag-driven movement’s network across the globe, but primarily in North America and Asia, from March to July 2024. It identifies key barriers to adoption, particularly for smaller nonprofits, and reveals significant gaps in A.I. policies and collaborative data agreements across the sector.
These findings come at a critical time as nonprofits continue to navigate the rapid development of A.I. and its potential impact on their work, says Woodrow Rosenbaum, chief data officer at GivingTuesday.
The hope is that “the nonprofit sector is not left behind in another technological revolution,” says Rosenbaum, and “the opportunity is that the nonprofit sector can actually lead the development of these tools in some really important ways.”
In the nearly two years since ChatGPT propelled A.I. into the public imagination, three distinct approaches to A.I. have emerged among nonprofits, the survey finds. More than half of nonprofits, which the report labels “A.I. Consumers,” have already embraced the technology and express eagerness to do even more with A.I. A little over a quarter of nonprofits — or “Late A.I. Adopters” — have yet to take the plunge but hope to do so in the future. Finally, 15 percent of nonprofits or “A.I. Skeptics” have tried A.I. but aren’t rushing to expand its use.
As nonprofits continue to adopt the technology, the report concludes, the vast majority will likely be using A.I. soon, regardless of organizational size. In fact, the tipping point for A.I. readiness appeared to be the hiring of a dedicated technology or data specialist, rather than a certain staff or revenue threshold.
“It’s not enough just to be big,” said Rosenbaum. “There’s real value in having data analysis expertise on your team, regardless of the size of your organization.”
While “A.I.” encompasses a wide range of technologies, the survey reveals clear trends in how nonprofits are currently leveraging these tools. Generative A.I., the advanced technology behind tools like OpenAI’s ChatGPT and DALL-E, stands out as the most widely adopted A.I. technology in the sector. However, nonprofits are also exploring other use cases, like A.I.-powered translation tools and data interpretation systems.
Despite the high adoption rate, the survey reveals significant doubts and barriers that nonprofits face in their A.I. journey. The study found a net comfort score of minus 16 on a minus 100 to plus100 scale, indicating substantial reservations about A.I. use, with data breaches topping nonprofits’ concerns, followed closely by worries about A.I. bias and potential job displacement.
When asked to weigh the balance of A.I.'s risk and reward, respondents were divided, with a slight skew toward reward and a significant portion still unsure about A.I.'s impact. Tech-savvy respondents tended to see A.I. as more risky than those less versed in the nuances of the technology.
Looking ahead, the survey reveals clear preferences among nonprofits for future A.I. tools. The biggest market gap identified is for A.I.-powered data organization tools that can help nonprofits better structure, clean, and manage their data.
Data analysis tools and virtual A.I. assistants also rank high on nonprofits’ wish lists. Interest is keen in ways to leverage A.I. to derive deeper insights from their data to streamline administrative tasks, freeing up human resources for more strategic, mission-focused work.
Mundane opportunities for advanced data analysis have been around for years, but the rise of user-friendly forms of A.I. has allowed nonprofits to rethink what their data can do for them and pave the way for better data practices, says Rosenbaum.
“We’ve needed to do more of that for the past 15 years,” he says. “So if this ends up being a lever to get organizations to do that a bit more, we’re going to see really broad positive outcomes.”
Other findings from the report:
- More established organizations (30-plus years old) were far more likely to be in the Late A.I. Adopters group, despite typically having larger staff.
- Education and health organizations are overrepresented in the group of A.I. Consumers, likely due to the longstanding use of data in these causes.
- About 70 percent of organizations have a data-use policy in place, but only 28 percent have been involved in adopting collaborative data agreements or guidelines.
- Nonprofits in the global south, particularly in India, show less concern about privacy and bias issues compared with those in the global north.
