What can generative AI do for your nonprofit?
Between the hype and the fear, is there a place for generative AI to support nonprofit operations and relieve some of the burden on overworked staff? I would answer with a qualified yes. Let’s dig into a few use cases after a quick review of generative AI.
For me, awareness of generative AI started in the fall of 2022 with ChatGPT. As someone who does a fair amount of professional writing, I was curious to see how AI might change my work. My early attempts were entertaining, but the output didn’t meet my standards and I recognized two important things that still guide my work when using these powerful tools.
The prompt matters, a lot. What you give these tools to work with will greatly influence the output.
AI tools can struggle with facts and accuracy. Often referred to as AI hallucinations.
A quick and silly example: I asked a generative AI tool to help me draft my wedding vows. The prompt I gave was to write vows using lyrics from the band Journey. The result was cheesy and clichéd, as you might expect. But the vows also included words from songs by Chicago. No harm was done because I wasn’t going to use this output, but it was an early reminder of a key limitation of these tools.
From that starting point, I found (professional) uses for generative AI that provided better output, and I was more prepared to understand the nature of editing and review required to get from prompt to final product.
Writing
One of the most common ways to tap into generative AI is for support with writing. There is a range of activities you can ask AI to help you with, and there are plenty of specialized tools for different types of writing.
Article headlines: You have written an article for your organization’s website and the only thing standing between you and publishing is a catchy headline. You can ask a tool like ChatGPT to create 5-10 options for headlines based on the topic of the post, or you can give the tool the full text to work with. While there might not be an outright winner from the AI generated options, you can probably find a few ideas to work with and refine them from there. Saving time to edit what generative AI produces is key to getting AI to work for you.
Content outlines: In this case, you know what you want to write about, but you are having trouble structuring your article or grant. You can ask AI to help in a couple of ways. First, you could ask it to provide an outline for the article. Typically it will give you a general structure to follow and you can decide if that aligns with your goals. Another approach: ask it to give you what questions people have about your topic. Getting specific about which people’s questions you want to consider, for example social workers, can help refine your output.
Editing: You can ask generative AI to provide feedback on your content. For example, if you are writing an opinion piece, ask AI to play the role of opinion editor at your target publication and offer feedback. Again, you will have to discern what feedback is helpful and what detracts from the point you are trying to make. If you are writing grants and need to shorten a response to fit a character or word count limit, generative AI can help with this as well. There are grant-specific AI tools that are better at this, but if you are using something like ChatGPT it is unlikely to get to your precise target. As with everything else AI, you will have to do the final edit to make sure that shortened response has kept your key points and met the character limitation.
I typically don’t recommend asking generative AI to draft full pieces, even if you are going to go back to edit them and add examples. The writing tends to be generic and unusually paced. Many readers will feel like something is off in the cadence but won’t be able to pinpoint it. AI also tends to overuse certain words. A couple words we have noticed in AI writing at AmplifyDMC include “unwavering” and “moreover.”
Meeting management
AI can take meeting notes for you. There are several tools that do this and they all have different strengths. Many have trial periods so you can test out which work best for your meetings. At Amplify, we like the tools that document what decisions were made and give assignments for action items.
Data analysis
Analyzing donor data using generative AI can enhance fundraising efficiency by predicting donation trends with higher accuracy and identify potential high-value donors who may have previously gone unnoticed. This predictive capability enables nonprofits to tailor their communication strategies and allocate resources more effectively, ensuring that efforts are focused where they are most likely to yield substantial results. For instance, AI can segment donors based on their giving patterns, interests, and engagement levels, allowing for personalized appeals that resonate on an individual level. This not only maximizes the potential for donations but also fosters stronger relationships with donors, enhancing their long-term commitment to the cause.
Ethics and Privacy
The potential for beneficial use of AI is great, but the current generation of tools have limitations that nonprofits should consider including privacy, data security, and bias. Managing privacy and data security are critical, as mishandling donor information can lead to significant trust erosion and legal repercussions. Nonprofits should implement robust data governance policies – not only for when data can be analyzed using AI, but overall practices for data storage, management, and use. Nonprofits can ensure that AI tools are being used ethically and effectively by being transparent about how and when they use AI tools to support their operations.
In addition, nonprofits should be vigilant in watching for bias in AI-generated content and even data analysis. There is an inherent risk of bias in AI models based on the information that was used to train them, which can negatively affect representations of racially and economically diverse people. While more attention is being paid to bias in these tools, there is little the average nonprofit can do to influence the base data that the tools were trained on. Just like with editing content to ensure that the tool didn’t introduce any errors, edit data and content outputs for bias to avoid perpetuating inequalities.
Is AI right for your organization?
Implementing AI in nonprofit operations should start with cautious optimism and thoughtful planning. Find a couple of use cases, like repurposing grant content for a newsletter or social media post (don’t forget to fact check and edit before sharing!) and try them several times. From there, you can assess whether AI is helping with that task or not. Then test generative AI’s capabilities with additional tasks until you find the right mix of support for your organization’s needs and goals.
Resources
Christopher Penn regularly uses his newsletter, Almost Timely News, to explore using AI more effectively. You can find it on his website or on YouTube.
Get better at writing prompts. While this post by Whitney Deterding on Spin Sucks is geared towards PR professionals, the recommendations can be translated to other uses.
Learn more about biases in AI tools and how they affect output and readers.
Next in Nonprofits recently hosted Submittable on their podcast to talk about responsible AI.