Are you currently working for an organization or prospect development (PD) shop that uses or is considering the use of artificial intelligence (AI)? Does your role involve data analytics from both the quantitative and qualitative sides of research? My experience over the last few years has been adapting to the rise of AI when explaining donor trends and moves management strategy to leadership. Suppose your job involves utilizing AI to shorten project timelines and communicating with other people. In that case, the webinar “Future-Proof Fundraising: Harnessing the Power of Generative AI” offers a lot of food for thought.

Here are five key points I took away from this insightful session, led by Cannon Brooke, founder of Brooke Solutions, earlier this year:

1. Data security should be considered! One thing to realize is that AI is learning from what we share with it. Brooke brought up how the accuracy of AI is something to consider along with surface ethical concerns that also arise with data security. For instance, OpenAI stores conversations for training purposes, which is something to consider in light of recent data breaches.
   
   Our donor information is confidential unless the donor grants us permission to use this data for reporting purposes. Confidentiality cannot be guaranteed when using AI, and organizations should establish best practices on how to mitigate potential data security risks.
2. AI can help summarize thoughts, but generating competent strategy… not so much. The more complex the needs, the less accurate AI will be in its output. As Brooke mentioned, we can use it to inspire the beginnings of ideas but we work in complex systems. For instance, moves management systems vary due to donor related variables. AI can give us the basics of a moves management system but is unlikely to accurately develop a strategy for how my shop should work with the data and donor knowledge.
3. The impact of prompting. From Brooke’s session I learned that there are varying levels of prompting strategy from zero-shot prompting to few-shot prompting. Even if you share deep organizational knowledge with an AI tool, the prompt you give will determine how accurate the output is. It’s like a chess game with AI, with the wrong prompts resulting in unsatisfying outputs. I think this conversation is a webinar in itself, and would be beneficial for small shops with limited resources.
4. Document analysis in seconds can be your best friend! As prospect researchers, we analyze so many documents and articles on potential and current prospects, that we may often lose track of the most important pieces of the story. I thought the part of the webinar discussing how we can review 990s and have AI do basic mathematical analysis on comma-separated values (CSVs) with hundreds of datapoints can be a time saver. While the old process was to use mathematical calculations in Excel or by calculator. Now, we can reduce the time of searching for financial and giving information of foundations for instance by attaching a CSV of prospect targets into an AI tool and prompt it to find the last five years of desired information.
5. Local large language models (LLMs) can be a good starting point. Prior to this webinar, I didn’t know downloadable private LLMs were available outside of creating one yourself. This could be a great way to practice prompting and to gauge how to improve the learning of the AI assistant before advocating for it to a larger audience at your organization. This option also gives prospect development professionals the ability to test which documents imported into the model have the maximum benefit.

Finally, I wanted to touch on the importance of this webinar for those who are new to AI and trying to better understand how to not only incorporate it into everyday tasks, but also get buy-in from leadership. Brooke had a very good opening about why we need to assess the pros and cons of this innovation. We must determine the ethics and accuracy of the output, or as Brooke put it, we’ll sound like a machine and not our own professional identity. This is crucial as prospect development professionals are tasked with more due to increased competition for fewer dollars. When we interact with the donors via omnichannel campaigns, it must sound like it came from us and what we express to leadership in analysis has to be as accurate as possible. This webinar did a great job teaching us the foundations of AI (like ChatGPT) while also going into the “why” which is the first stage in technological adoption.