The Organizer #75 | Fundraising

What do I need to know about fundraising in 2024? Be aware of AI and ChatGPT in social impact work

AI in social impact work

By writing this article we are breaking a promise we made to ourselves. We said we’d never write about this. There is no Venn diagram of people like me and people who bang on about how awesome AI is and what a game-changer it will be. There are just two circles, sitting in isolation, side by side. But, AI in social impact work is inescapable, so here we are.

Technology extends our capabilities and impact. But, I wouldn’t recommend AI to most nonprofits. It’s an example of making your work “easier” in a way that does not make it better. Anything that promises the moon should be treated with skepticism, especially if it comes at the expense of quality.

That said, if you are involved in social impact work in any way shape or form, you need to know AI basics.

AI is not the future of social impact work – it is the present. AI is everywhere, including inside your own organization. Because it’s everywhere, because it has spread so quickly, and because the people who embrace it often champion it fully and completely, it’s hard to get a handle on the key bits of information that nonprofit managers really need to know.

This article is written specifically for the people who don’t want to read it. We went there, so you don’t have to.

The key things you need to know about AI in social impact work:

1) “AI” or “Artificial Intelligence” almost always means ChatGPT.

You don’t need to understand AI, but you do need to know how ChatGPT works. It’s an interface similar to a search engine: you ask a question or type in a prompt, and ChatGPT spits out an answer. Unlike a search engine, it doesn’t provide a list of options with links to pages that might answer your question. It provides a single, complete, confident response.

The key thing to know is that ChatGPT doesn’t have the ability to vet information for accuracy; instead, it is trained using sources all over the web to generate an answer that looks accurate (this is why it can ‘hallucinate’).

If you’ve never tried ChatGPT, you can access it free here and play along with some of the examples below.

Just remember that the walls have ears. You are adding additional information to the pool of training data every time you create a new prompt; don’t share any confidential or proprietary information.

2) Your team is using ChatGPT to write grant proposals and fundraising appeals.

AI can be used for most aspects of social impact work. According to this glowing article, ChatGPT makes it easy to write a proposal. Here’s an example of an actual grant proposal we “wrote” in mere seconds.

If you’ve never seen a real-world example before, have a peek: ChatGPT is better than you might think at structuring a grant request.

People are also using it to write their email fundraising appeals. For instance, this example of a personal email with a donation request.

3) Your team is using ChatGPT to edit their fundraising appeals.

AI tools don’t just write for you or answer questions (unlike search engines, ChatGPT does not directly pull from sources). You can feed in context when you ask a question, like this:

‘Give me feedback on my response to Grant Application X using assessment criteria Y.

Since ChatGPT is designed to assess what would be ‘expected’ given your prompt), the AI tool will tell you how “well” you’ve answered the question and provide suggestions for improving your response.

You can see a prompt example here.

4) Foundations are using AI tools to screen for ChatGPT-generated applications.

In the same way that ChatGPT can know what a grant proposal should look like, tools like GPTZero allow you to test content to see if it was written by a human. It is important to note some foundations are automatically eliminating AI-generated applications from consideration given concerns about accuracy of the information.  

5) Funders are also using AI to read and rank, and rate your grant proposals.

Grant management software now comes equipped with AI-informed tools to screen and rank applications, a time-saving feature that software providers see as a benefit.

Here’s an example from SmartSimple Cloud +AI.

6) Communications folks and spokespeople are using ChatGPT to write speeches and remarks for fundraising events.

Does this sound familiar?

7) Fundraising and marketing staff are using ChatGPT to generate strategy documents they need to pitch your work.

Yes, robots can write a theory of change.

Yes, they can write a project budget.

8) People increasingly use ChatGPT in place of search engines.

This means your team may be using it to answer science and economics questions or to provide factual information that would normally take hours of research and expertise.

For instance, these prompts show how it generates responses to important policy questions:

To be super clear, ChatGPT does not have access to current information, cannot discern the quality of information it provides, and reflects all kinds of biases. The fact that OpenAI doesn’t share where it gets its training data will be a big shiny flag for most social impact researchers and advocates.

We want you to know that people are using it like a search engine, but we do not endorse this use for ChatGPT.

So what?

Yes, ChatGPT and similar AI tools make time-consuming aspects of social impact work faster. Yes, they give you feedback and access to general information quickly. They can create a rough first draft or provide a model that you can learn from. The results are also riddled with errors, misinformation, and (in the case of unchecked copy or requests) commitments that your organization may not be prepared to fulfil.  

AI tools can make you fit in, but they can’t make you stand out. They don’t do original research or thought, and they don’t know your donors or programs, especially if you work on society’s fringes. They can’t solve unsolved problems. And they definitely don’t get facts right. Those are all things that social impact organizations exist to do.

The very strength of ChatGPT — speed — can be a weakness in social impact work. Trust, relationships and insights are critical to impact and important parts of fundraising; they are built with time and experience. Impact and knowledge comes from contact with our communities, while shortcuts dissolve trust.  

And yet…

ChatGPT and AI tools are everywhere, and just emerging.

It’s not safe to turn up our noses and tune it out. The promise of being able to do work quickly and with little effort is too alluring for most to resist. The value of being able to see examples, templates, and get immediate feedback in a sector where training and manager attention are limited is extremely appealing.

AI in social impact work isn’t a technology issue for technology people – it’s a skills, organizational culture, and training issue. It’s a knowledge-sharing issue. It’s a community-accountability and engagement issue.

You don’t need to know the tech inside out. Just know it’s here, and be prepared to guide your organization’s outreach efforts in ways that fit with your long term vision and values.


The Organizer is a newsletter for people working to create equitable and sustainable communities. Whether you are part of a nonprofit, a charity, or a social enterprise, this newsletter is for you.

Each edition, we explore one aspect of social impact work. We answer a common “How do I …?” question, and we tell you about a tool that will help make your work a little easier. Subscribe for free at Entremission.com.