When it comes to technology adoption, nonprofits tend to trail enterprises. They often lack the money and technical know-how to make the most of the latest innovations, even as they have the most to gain from them. We asked Kevin Barenblat, the co-founder of Fast Forward, a nonprofit that helps other nonprofits use technology for social good, how organizations are leveraging AI to advance their missions.
How are nonprofits evolving in terms of their AI use?
The biggest shift that we’ve seen in AI in the past few years is the rise of the conversational AI that started with the launch of ChatGPT two and a half years ago. Throughout all of computer history, you had to understand the language of the computer in order to interact and get the data you want. One of the things that’s changed in the past couple of years is now we can use English, so it’s more accessible to more people to have computers help us do the things we want to do.
Now, we’re seeing more nonprofits using AI to scale their impact and solve humanity’s most urgent issues — we call these organizations AI-powered nonprofits. WattTime is analyzing satellite imagery to understand global climate emissions. Quill is using AI to guide students to become better writers and critical thinkers. RebootRx is using AI to identify generic drugs that can be used in treating cancer. What started as almost a toy, where we could get ChatGPT to write a poem for us about our friends for their birthday, now is being used in enterprises and in nonprofits to do real work.
What are you seeing as the biggest opportunities for nonprofits as they experiment with AI?
The way AI used to work is you would train a model to do something very specific. Quill, for example, would take a paragraph and build a model specifically for that paragraph on teaching students how to engage with it, and it taught the model how to react to different responses from students. The AI was trained specifically on that model. And now it’s becoming possible for AI to do more complex tasks by stitching together those individual things. So, the hope is that it’s more useful for a wide variety of problems and tasks.
Another AI-powered nonprofit, Digital Green started by going to farms and showing videos in the local language to help farmers improve their farming practices. Now it provides them much more customized advice from a mobile device that’s intelligent and works real time. So, I think we’ll see that in a lot of places.
This is the dream of the AI optimists — for AI to improve our global well-being. In Fast Forward’s new 2025 AI for Humanity Report, we found that 84% of AI-powered nonprofit respondents said funding would most help them further develop and scale AI. This is important because the data shows a clear relationship between resources and reach. At the smallest budgets, AI-powered nonprofits are serving thousands, a median of just under 2,000 lives. By the time budgets cross $1 million, median reach jumps to half a million people. And at more than $5 million, AI-powered nonprofits impact a median of 7 million lives.
How are nonprofits accessing these tools? Are they building them in-house or using off-the-shelf solutions?
We see a mix. Most AI-powered nonprofits start with off-the-shelf tools to get up and running quickly, then move to in-house solutions once they’ve proven that the approach is worth deeper investment. For example, Digital Green initially piloted its farmer chat feature using an existing platform, and later built its own version.
Even when using pre-made tools, most nonprofits customize them to meet the unique needs of their communities. Ultimately, they don’t focus on AI for its own sake — they care about the impact it has.
Nonprofits are problem-driven: they start with a specific challenge and choose the technology that best helps them solve it. Unlike tech companies, which often aim to build broadly scalable products, nonprofits focus on solutions that fit their mission and users.
Ethics, privacy, and bias are also top of mind. According to our research, 61% of surveyed AI-powered nonprofits customize large language models with their own data to better serve their communities, and 70% regularly incorporate community feedback into system updates.