Ed. note: Throughout the past 18 months, the Goldhirsh Foundation and its LA2050 initiative have provided complimentary AI (artificial intelligence) consulting, training, and workshops to the Los Angeles impact community. The work is led by Jen García, Goldhirsh Foundation’s AI EIR (executive-in-residence). Below – and in other posts – García shares practical insights, real-world use cases, and emerging research to support nonprofits’ responsible and ethical adoption of AI tools.

Earlier this summer, Goldhirsh Foundation served as a partner in OpenAI’s national AI Nonprofit Jam. Participants included the leadership of many LA2050 grantees, along with other nonprofit organizations serving the people of the region. [Read more.] I served as a tech mentor that day with other AI experts from companies like OpenAI and Emerson Collective.

To help inspire those assembled, Goldhirsh Foundation president Tara Roth spoke, along with others including Jade Clemons from AltaSea at the Port of Los Angeles (an LA2050 Grants Challenge winner), and Jason Frazier of The Music Center. Clemons and Frazier each made presentations about how their organizations are leveraging OpenAI’s ChatGPT for productivity.

The day was nothing short of thrilling: watching attendees from a range of AI usage abilities learn and share with one another was exactly what OpenAI and Goldhirsh Foundation teams had envisioned. As I introduced myself to each table of attendees, they shared what they hoped to leave with, and I knew they were in for a treat: AI tools, particularly ChatGPT, can enhance nonprofits’ productivity, allowing them to focus on serving the community. One goal of the Jam was for attendees to leave with a project they could take back to their organizations.

Quan Huynh was a Jam participant. He’s the executive director of Defy Ventures Southern California – an LA2050 Grants Challenge winner – and we’ve been providing him AI training and consulting. Huynh created a role-playing coach to help prepare his fellow executive directors around the country for donor meetings. Like many executive directors of nonprofits, Huynh and his colleagues are pressed for time, so access to a custom GPT that could not only automate research about the target donor but also prepare them for the meeting by role-playing the interaction is highly beneficial with clear return on investment.

“Before [using this custom GPT], you would throw in Google search,” Huynh told me in a call after the Jam. “But now there's something that can actually automate it, talk to you, and give you feedback.”

Since creating the bot, which he called Donor Dialogue Coach, Huynh has used it to prepare for a meeting in which he requested a $30,000 grant – and won it. When I asked him for words of advice for fellow nonprofit leaders, he encouraged others to turn a key question on its head: “Don’t ask for it to be built for you, simply start building it yourself.”

At the Jam, one of the attendees asked me about the differences between a custom GPT and a Project in OpenAI’s ChatGPT. I explained to them: GPT is short for “Generative Pre-trained Transformer.” “Generative,” because these large language models generate new content, “pre-trained,” because they were trained on vast amounts of data before being fine-tuned, and “transformer” because they use transformer architecture designed for understanding and producing language.

Custom GPTs are user-tailored versions of ChatGPT built for specific tasks, whereas Projects organize related conversation threads. Both can be customized with specialized instructions and files, but only GPTs can be shared via link with either the public, anyone with the link, or only within your organization (using a Team or Enterprise account). GPTs are highly customizable, allowing users to program the GPT to automatically utilize a variety of tools like web search, Canvas, image generation, code interpreter, and data analysis functions. Although conversations within Projects can do so too, one must manually select or directly prompt the usage of these tools.

Huynh’s creation of a Donor Dialogue Coach is a perfect use of a custom GPT due to its shareable nature and dedicated interface, allowing the user to chat with it just like they would if interacting with ChatGPT alone. And, if he wanted his research on the target donors to be organized, he might choose to utilize ChatGPT’s Deep Research and place all related chat threads into a dedicated folder that he could name as he wishes. Within the Project, he would see all the related conversations in sequential date list format.

Other nonprofit-related custom GPT use cases include grant writing support, new hire onboarding assistance, or an organization policy support wherein staff can easily query policy documents to get instant answers and accompanying citations. The possibilities are vast, and limited mostly to the user’s imagination.

Photo: Tara Roth, president of the Goldhirsh Foundation, with Jen García, Goldhirsh Foundation AI executive-in-residence and author of this blog post, at the AI Nonprofit Jam held in Los Angeles.