Changemaker's Guide to Generative AI

Whether you’re organizing a park cleanup, launching a school supply drive, or helping your friends learn about mental health, large language models (LLMs) can be your secret weapon.

Think of LLMs as the coach, planner, designer, and translator in your pocket. They don’t replace your voice or passion but they do make it easier to go from “I want to help” to “Look what I just made.” Of course, LLMs can have downsides. We’ll talk about those, too.

Nonetheless, they’re powerful tools! Here’s how to team up with AI to boost your impact in your own neighborhood.

You know your community best. Sometimes the hardest part is knowing where to start. LLMs can help you connect your passions to local possibilities by asking smart questions, surfacing issue areas, and suggesting creative actions.

Iterate until you have responses you like, and then ask the LLM to generate a full action plan for your favored action.

Once you’ve got the idea, it’s time to plan your project. How might you best go about delivering upon the dream you have for your community?

Text-based LLMs can help you write everything from helping craft your project plan to crafting specific planning and outreach materials.

You don’t need Photoshop skills to make something that pops. LLMs can help you write design briefs for tools like Canva, generate layout ideas, and craft image alt text for accessibility.

With LLMs, creating prototypes of websites and other technical applications is more accessible than ever. Just by telling the AI what you want your website to look like, it can generate code for you to share with others or even publish to the internet!

Websites can be a great way to share information and spread the word on causes you care about. For example, you could make a website that shares key information about climate change, or that promotes a volunteer opportunity.

Or, maybe you have a bigger vision. LLMs can help you write code for simple tools—like digital sign-up sheets, resource maps, or mini web apps that help your action run smoothly. Or, you can use LLMs to write code that analyzes community data. Its vibe coding meets actual engineering, and we’ve been testing this pathway extensively.

Here are some tips:

• Be specific about what you’re looking for. In most instances, you won’t be able to support a major enterprise application. Make sure the LLM knows you need simple, streamlined code with a simple application framework.
• Be specific about what you want to do with the code. For example, if you’re just looking to test the UI and UX with people in your community, you can create simple webpages that mock a larger dynamic experience.

Making your action accessible isn’t optional—it’s part of good leadership. LLMs can help you rewrite materials in simpler language, translate them to other languages, and add accessibility features.

Of course, large language models don’t know everything, and they can add in some new problems, too.

Insight: LLMs sure sound like they’re smart!

Large language models certainly sound like they know their stuff, don’t they? Well, they don’t! They’re synthesizing output - text, usually - that its algorithm has determined seems like it should fit. So, a language model is apt to tap outdated information, make up resources or references and otherwise hallucinate. Of course, it doesn’t know it’s doing so; it’s just trying to provide a helpful answer. The answer might just be wrong!

Insight: LLMs can be biased!

Language models are trained on extraordinarily large datasets - that is, the text of the internet. But the text of the internet itself can have bias; and a language model can over-anchor on biased versions of text. AI creators also, intentionally and unintentionally, insert their own biases when developing new models. Therefore, the response you receive may reflect that same bias. Treat responses with skepticism.

Insight: LLMs don’t have emotions.

An LLM doesn’t know what it is to care, and can’t sense who it’s supporting. You or your audience may have unique needs, including accessibility needs, that it simply can’t understand. Likewise, it doesn’t understand what your community is like, or what resources you may or may not have access to. Be sure to review all outputs and shape them to suit your needs!

You don’t need to be a coder, designer, or organizer to make a difference. You just need an idea—and maybe a little help bringing it to life. Whether you’re just getting started or planning something big, AI tools like ChatGPT are here to help you build confidence, creativity, and impact.

Want to see where you could start? Check out the Choose Your Own (Service) Adventure action to find your cause and take your first step.