Nonprofits are well-known for being able to do a great deal with minimal resources. With this in mind, the incorporation of Large Language Models (LLMs) like ChatGPT and Bing Chat into their toolbox seems like a natural next step.

Research regarding the impact of AI is still in its infancy, but early studies show that far from replacing human workers, LLMs like ChatGPT and Bing Chat empower workers.

Why You Shouldn’t Fear AI: Insights from Early Research

In a recent study444 college-educated mid-level professionals from various occupations, such as marketers, grant writers, consultants, data analysts, human resource professionals, and managers, were asked to complete tasks. Half of them were randomly exposed to ChatGPT during the process.

The tasks in the study consisted of 20-to-30-minute assignments, including writing press releases, short reports, analysis plans, and delicate emails. These tasks were designed to resemble real work performed in the participants’ occupations. The participants reported having completed similar tasks before and rated the assigned tasks as realistic representations of their everyday work.

Impact on productivity

Workers exposed to ChatGPT completed tasks 37% faster, and the resulting text quality was significantly higher. This improvement seems to stem from workers spending less time on drafting and more time on editing and polishing their work.

productivity and time changes

Without ChatGPT access, people dedicated half their time to creating rough drafts, whereas those with access spent over half their time editing. Furthermore, ChatGPT promoted egalitarianism by compressing the quality and time spent on tasks, reducing the gap between the most efficient and skilled writers and the rest. This decrease in inequality resulted from lower-scoring participants benefiting more from ChatGPT access

Impact on job satisfaction

Following each task, participants were asked about their enjoyment and perceived skill level. In the study, ChatGPT access notably boosted job satisfaction and workers’ sense of self-sufficiency. It also fueled optimism towards LLMs and AI technology in general, with workers becoming more enthusiastic and increasingly expecting AI to positively impact the world and their lives.

job satisfaction

How to Use LLMs at Your Nonprofit

Turning ideas into drafts

You may have heard stories about how AIs like ChatGPT and Bing Chat can be inaccurate and generally unreliable when precise facts are required. However, these concerns overlook the bigger picture. The true strength of these AI models lies not in writing entire pieces autonomously, but in how they can enhance your performance.

Consider your AI companion as an engine that transforms bullet points into paragraphs.

For instance, let’s say you’re working on a report for a nonprofit called “Animal Shelter,” which rescued over 5,000 street animals and found homes for 3,000 of them in the past year.

If you request the AI to compose the introduction to your report using only the nonprofit’s name, be prepared for the model to produce a rather generic response:

Prompt: Write the introduction for the 2022 annual report of the nonprofit organization Animal Shelter.

Output: We are pleased to present to you our 2022 yearly report, highlighting our efforts and accomplishments throughout the year. As a nonprofit organization, we remain committed to our mission of providing a safe and caring environment for animals in need, as well as educating and engaging the public on animal welfare issues. This past year has been one of significant challenges and changes, with the ongoing COVID-19 pandemic affecting all aspects of our operations. Despite these obstacles, we have remained steadfast in our commitment to our cause, and we are proud to report that we have made great strides in advancing our mission.

While it’s not a bad attempt, it doesn’t address the specifics of the organization’s achievements. Instead, see what happens when you supply the model with the relevant facts yourself:

Prompt: Write the introductory paragraph of a report for the following nonprofit: Animal Shelter. During 2022, it achieved the following::

*Rescued over 5000 street animals
*Helped more than 3000 find a new loving home

Output: We are thrilled to present the 2022 report for the Animal Shelter, a nonprofit organization dedicated to rescuing and providing shelter for animals in need. This past year has been one of tremendous growth and success for our organization, as we worked tirelessly to rescue over 5000 street animals and help more than 3000 of them find a new loving home. Through the tireless work of our staff, volunteers, and supporters, we have made significant progress in achieving our mission of providing a safe and caring environment for animals in need. In this report, we will detail our rescue and adoption efforts, medical care services, education and outreach programs, and fundraising initiatives, all of which have enabled us to make a positive impact on the lives of animals and the broader community. We are proud of the achievements of the Animal Shelter in 2022, and we look forward to continuing our important work in the years to come.

Prompt Engineering Tips

Prompt engineering is the art of asking the right questions or providing the right instructions to an AI language model to achieve optimal results. Effective prompt engineering takes into consideration factors such as specificity, context, target audience, and the desired format of the output. It often involves iterative tweaking and experimentation to find the best way to communicate the user’s intent and requirements to the AI.

Always strive for specificity! The facts, numbers, and ideas are up to you! You wouldn’t expect a person to get the facts about your organization right or understand what you prioritize if you never explain it to them.  Consider these AI models as a highly efficient writers from outside your organization;  tell them what you want and they’ll mostly get it right.

Specificity isn’t limited to facts, numbers, or ideas. Consider your desired writing style, length, keywords, and any stylistic guidelines to adhere to. Here are some additional prompt engineering tips:

  • Include who your target audience is in your main prompt  
  • List the main points of your article as bullet points and ask the LLM to write one or two paragraphs for each bullet point
  • Iteration is key to getting the most out of LLMs. Use the  outputs from your previous prompt as part of your next prompt. For example, ask the LLM to make it more engaging, shorten it, etc.  
  • Ask for multiple options, instead of asking for a title based on a paragraph, ask the LLM to write you 5 potential titles
  •  LLMs excel at grammar and syntax, so ask them to read your text and correct it. They work wonders as a second pair of eyes.

Conclusion

In a nutshell, embracing AIs like ChatGPT and Bing Chat can revolutionize productivity and improve job satisfaction in nonprofits. By applying basic prompt engineering tips, organizations can transform their ideas into engaging content more efficiently. As AI keeps advancing, its value in the nonprofit world will only grow. I hope you found this article useful and inspiring. I’m very excited to see what nonprofits will be able to accomplish with this new set of tools.

 

 


Other articles picked for you