A beginner’s guide to AI in annual giving

1. Sophistication of output

• Output of the new AIs is indistinguishable from text, speech, images, etc. produced by intelligent, skilled, and highly knowledgeable humans.

2. Natural language processing

• The new AIs can understand questions posed to them in casual, everyday language and can respond in kind. This lets people make productive use of them with little training.

3. Advanced build features

• Ability to self-optimize
• Incorporation of deep-learning neural networks
• Use of automated statistical analysis
• Training on hugely massive data sets

One of the most common early uses of AI is content creation. That includes general writing assistance—using tools like ChatGPT to help with the clarity, tone, grammatical correctness, conciseness etc. of various kinds of writing—as well as the creation of deliverable types that depend on some level of specialized knowledge—budgets or strategic plans, for example. In the latter case, AI not only helps with the mechanics of writing but also supplements users’ subject matter expertise. This technology can provide immense value to advancement teams, especially those with limited bandwidth, to be more efficient with their time and efforts.

Say you want to develop a new email series for your Calendar Year-End campaign. You can use a tool like ChatGPT and provide it with your campaign plan (donor segments, fundraising goals, institution information, etc.) along with prompts for the email series (topics you want to cover, parameters you want it to follow, length and content to include, etc.) and it will draft the emails series within seconds. Remember, AI generates this written content based off the data you feed it. So, the higher quantity and quality of historical annual giving campaign content you can feed it, the better. Be careful— common challenges with generative AI are that it writes in general platitudes, can have inaccuracies, and is often missing a human spark. This is where it’s critical to bring in industry professionals to edit and supplement AI’s content creation.

Other early uses of AI:

• General writing support (i.e., tailoring marketing copy to different audiences, checking copy for grammar, drafting marketing copy)
• Writing and content creation for specific deliverable types (i.e., creating budgets, drafting job descriptions)
• Data analytics (i.e., identifying triggers most likely to convert prospective donors)
• Brainstorming
• Communications automation

As you and your team begin to explore how you can use AI in your annual giving work, you will likely first go to ChatGPT or another similar free model. Here are some things to keep in mind when starting out.

1. ChatGPT generates output based on prompts you give it; the usefulness of its answers depends on the quality of the instructions you provide.
2. You will never get what you need from ChatGPT on the first try; you’ll need to iterate, tweaking your prompts based on its responses.
3. There’s really no way to get good at prompting ChatGPT except through practice (though trainings are definitely helpful). While ChatGPT’s output is typically very good, it invariably includes inaccuracies or other problems; always check its output before using it.
4. It is no substitute for subject matter expertise. Because ChatGPT output needs to be checked for accuracy, it is of limited use for tasks or subjects about which a user is not knowledgeable.

As AI progresses in its accessibility and capabilities, it’s important to have a general understanding of how it can benefit your team in the long-term. First, as long as AI’s work needs to be checked for errors before being used, you’ll need someone with the requisite level of subject matter expertise to use it.

Secondly, AI promises to greatly improve the productivity of your team, a benefit that is especially appealing given that advancement offices around the country continue to be understaffed.

Even though AI cannot substitute for subject-matter expertise, it can make your existing experts more productive and help staff develop new areas of expertise. Remember:

• You’ll still need knowledgeable humans to check and troubleshoot output from AIs (marketing copy, computer code, etc.)
• ChatGPT and other AIs can drastically reduce the amount of time that it takes even highly expert staff to perform core tasks.
• ChatGPT is an infinitely patient and informed coach and tutor, ideally suited to helping staff acquire new skills and knowledge.

Navigating data quality and governance poses a significant challenge for many advancement offices, with partners frequently expressing the complexities of data collection, utilization, and storage. The struggle to manage data effectively in-house becomes a critical consideration, particularly when contemplating the integration of AI.

Take steps to start cleaning your data

You can take steps to start cleaning the data at your institution. Here are a few things to consider:

1. Do you have one centralized location where all your data conglomerates?
2. Where do you collect data for your system? Consider both internal departments (ie. Student Affairs, Finance, Institutional Research, Schools and Colleges, etc.) and external parties (ie. third-party software, community information, etc.).
3. Are your gift officers consistently inputting visit notes, likes and dislikes of donors, conversations about campus with prospects, the feelings of alumni and donors about situations on campus? If your answer is no, start there.

AI’s potential is undeniable, but privacy concerns are legitimate. When you upload data to a software or platform not owned or licensed by your institution, you run the risk of sharing personal data which could be stolen or used inappropriately. As ChatGPT and other generative AI platforms develop and mature, surely there will be a larger focus on data privacy and protection. But for now, it’s crucial you are aware of the privacy risks, especially when using sensitive donor data. And be sure to only use AI-driven technology with partners and software you trust. A good rule of thumb for now is to assume that anything you put into ChatGPT may eventually be accessed by the general public. Don’t enter anything that you would not be comfortable posting on your institution’s website.

As mentioned, AI needs human expertise to help optimize and manage its output. It’s important to remember that AI cannot replace humans, but rather should complement and support the work we do in annual giving. You will get the highest performing results out of your AI experience by having advancement professionals who review donor lists, edit content, and ensure it is being fed with smart prompts. However, many advancement teams are already struggling with staff bandwidth and struggling to hire and retain experts in the advancement industry.