# Nonprofit AI: LLM 101 and Best Practices Roundup

An episode of [Community IT Innovators Nonprofit Technology Topics](https://invisiblenetwork.ai/podcast/1536843269), 8 Sep 2026, 42 min.

Audio: https://www.buzzsprout.com/1029925/episodes/19771987-nonprofit-ai-llm-101-and-best-practices-roundup.mp3

Carolyn Woodard covers two things this week: a quick and basic explainer of how large language models actually work under the hood, and a roundup of the best practices for using AI at nonprofits that Community IT has built up over nine months of this podcast and 3 years of our monthly webinars. First, the basics. An LLM is not a database and it is not doing a Google search. It is predicting the next most likely word, or token, based on patterns learned from the huge amount of text it was trained on. Image and video tools work differently, using a diffusion model that refines random noise into a picture, which is why video generation costs roughly 2,000 times more energy at the data center than a text prompt. "Hallucination" or being confidently wrong is a feature, not an AI bug. OpenAI research from last year shows it is a predictable result of how models are trained to guess a fluent-sounding answer rather than say I do not know. Then Carolyn walks through the best practices Community IT recommends for any nonprofit using AI, from naming an AI owner to keeping a human editor in the loop on anything that goes out under your name. The full version, with links to Community IT's past episodes and resources, lives in our AI for Nonprofits guide on the website. This episode covers: • What an LLM is actually doing when you type in a prompt, and why that mechanism is also why LLMs are bad at math. • What the fact that image and video generation use so much more energy than text means for your organization's AI choices. • Why hallucination happens, and how to prompt AI to tell you when it is uncertain instead of guessing. • A dozen or so of Community IT's best practices for nonprofit AI adoption: talk about it openly and often, name an AI owner, build hands-on literacy, verify what you can verify, keep a human editor in the loop, and many more • Why evaluating AI tools against your organization's values, not just their features, matters. Resources Mentioned: • Running Your Own LLM, Nonprofit AI Podcast • AI for Nonprofits Guide, Community IT • Values-Aligned AI Decision-Making, Community IT • AI Acceptable Use Policy Template, Community IT • Why Language Models Hallucinate, OpenAI, 2025 • Nonprofit IT Management Reddit Community New every Tuesday. _______________________________ Start a conversation :) • Register to attend a webinar in real time, and find all past transcripts at https://communityit.com/webinars/ • email Carolyn at cwoodard@communityit.com • on LinkedIn • on reddit/r/nonprofitITmanagement • on the Community IT website Thanks for listening.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.