The advent of generative artificial intelligence has been a gamechanger from what began in the 1950s as a tool for working with simple datasets. With new capabilities emerging every few weeks, developers are continuing to find new ways to apply and scale the technology to benefit nonprofits.
How to get started, where to go, and what it will cost were among the topics addressed by three expert panelists during a webinar hosted last week by The NonProfit Times and multinational technology and business services provider SAP.
“AI not only means artificial intelligence. It can also stand for altruistic innovation,” said Timo Elliott, global information evangelist for SAP. As you think about how your own organization will benefit, it can also help to think of AI as an awesome interface, as a tool for gaining advanced insights, and as a vehicle for accelerated implementation and automation of processes you’re already using. Some of these capabilities overlap. To illustrate, the panelists offered the following breakdown:
- Awesome interface: “Nonprofits interact with a lot of different stakeholders, including people who may not have computer skills or even know how to read and write,” said Sean Kask, chief AI strategy officer for SAP. Chatbots based on large language models get around this by enabling natural human-like interactions, whether it’s from a person seeking services or a prospective donor seeking easy access to information buried deep in an organizational report they’d rather not wade through. Your own people working on-site or in the field can also use it to quickly query the system about information pertaining to finances, volunteer management, grants management, and more. The same technology makes it possible to upload a PDF document or spreadsheet into the software and immediately begin querying it regarding data contained in the report.
- Advanced insights: AI algorithms can permutate and rearrange data from a near-infinite number of datasets to place usable information at your fingertips. Mapping out the distribution of food aid from the UN’s World Food Program recently became much simpler as a result, to cite one example offered during the webinar. With predictive modeling, algorithms can also rank prospective donors based on how likely they are to contribute during a future fundraising appeal in the same way they help businesses optimize marketing campaigns by sorting and identifying prospective customers. “If you think about commercial leads and apply it in a nonprofit sense, we can scale that out to get the same insights. It’s the same math and same underlying processes, just applied slightly differently,” said Jared Coyle, head of SAP’s Creative Collective subsidiary.
- Accelerated implementation: Many nonprofits are already embracing automation to streamline tasks such as grant writing. Increasingly, these automation capabilities can be applied to other business operations such as sorting, classifying, and forwarding incoming emails according to the type of request. The same holds true for processing travel and expense reimbursement forms submitted by employees electronically as PDFs. “We all know those people who really know their way around an Excel spreadsheet and have created their own little workflows and applications using simple Office tools,” said Elliott. “If I have one bit of advice, it would be to seek out those people and get them interested in experimenting with these new tools because there’s so much more they can do now than before. It’s different from the previous technology, and it’s very accessible.”
The headlines are full of amazing examples of what AI can do today and the additional things it will be able to do tomorrow. Yet, said Elliott, “the best way to get started with AI might be to use it to automate some of the so-called ‘boring’ things you’re already doing now.”
Ongoing advancements in AI will likely bring lower costs as well. “It is going to get cheaper, and I think that is critical from the standpoint of making sure your costs are reflective of the value you’re providing to your donors,” Coyle said.
Nonprofit leaders whose organizations lack the resources to hire their own coding experts or build their own software might still be hesitant to take the plunge. But with code generation capabilities increasingly embedded in AI technology, growing numbers of them no longer will have to. “The good news with a lot of these Generative AI models is it flips on its head the way AI works,” said Kask. “Before, you would have to train the model to do what you wanted it to do. With a Generative AI model, the heavy lifting of training the model has already been done.”
Many large language models in the generative AI space come with free-to-try options, the most well-known being ChatGPT developed by Open AI. Another is Anthropic Claude. Using these free versions can be a great way to get your feet wet. However, the panelists cautioned against using them for grant writing or other functions that might expose proprietary information.
Aside from the obvious risks of doing so, this has also resulted in the embarrassment of nearly identical grant applications being submitted by different nonprofits, in some cases with the names of competitor organizations still listed because someone neglected to remove or correct the information from previously used templates. “If you’re using the free version, they’re probably storing your data. That’s how they improve it,” said Elliott.
Bias is another risk of using AI software that hasn’t been fully tested or vetted, as for example when using it to assess eligibility for services or to write performance reviews. “Your reputation is what you live off of,” said Kask. “If you’re using AI to match invoices to bank statements, it’s probably not a concern. But if you’re using it to predict prison recidivism rates and those kinds of things, you cannot afford any missteps.”
Several free online courses are available that delve further into these topics. AI for Everyone, taught by AI pioneer Andrew Ng, is one that is recommended by Elliott. It consists of four modules totaling five hours and is offered by Coursera.
