Generative AI

Table of Contents

Generative Artificial Intelligence (Generative AI, often shortened to just “AI”) is a set of tools that can help you create, organize, and share information more effectively. At UMaine Extension, you can use AI to support education, research, and outreach by saving time, generating new ideas, and improving accessibility.

This page is for all staff, whether you’re completely new to AI or already using it. Even if you choose not to use AI in your own work, these technologies are already built into many of the tools, systems, and information you interact with every day.

Along with guidance, support, and resources for using tools like Google Gemini and ChatGPT, our goal is to help you build a clear understanding of AI and its potential risks. This knowledge will help you spot where AI is at work, ask the right questions, and make informed decisions in your day-to-day work.

Extension Generative AI Guidelines

UMaine Extension’s Generative AI Guidelines sets expectations for when and how staff may use AI tools including Gemini and ChatGPT, and embedded AI features in other systems. AI can be used to save time and improve communication, but it must never replace human expertise. Every AI output must be reviewed, verified, and edited to match Extension’s research-based, relationship-centered, and community-engaged mission.

Generative AI Guidance Summary

Extension AI use should:

  • Be intentional, not automatic.
  • Be accurate, fact-checked, and transparent about AI’s role.
  • Maintain Extension’s tone, quality, and integrity.
  • Never use personal, confidential, or sensitive information.
  • Consider environmental impact and only use AI when it adds value.
  • Be informed by continual learning about AI’s risks and capabilities.

What’s allowed:

  • Internal use: You can use AI to make your own work easier and faster for internal purposes.
  • External use (with SME review): You can create public-facing content only if a subject matter expert checks it for accuracy and tone.

Automating interactions with the public (e.g., AI-enabled chatbots, web tools or interactive calculators) requires prior Extension Leadership Team approval with a formal quality assurance monitoring plan.

What’s not allowed:

  • No confidential or personal data: AI tools do not guarantee privacy.
  • No realistic, lifelike images or video: These could be mistaken for real images and cause confusion.
  • No high-impact automated decisions: AI cannot replace human decision-making in areas like hiring or evaluations.
  • No harmful or copyrighted material: Avoid generating anything that could harm, mislead, or violate rights.
  • No AI in meetings without participant permission: Inform all participants and get explicit consent before using AI tools to record, transcribe, or generate content from meetings.

Disclosure:

If AI significantly shaped the ideas, structure, analysis, or creative direction, let your audience know. Minor edits, formatting help, or small idea prompts don’t require disclosure. You can use your own wording. One example is, “Created by UMaine Extension staff with generative AI assistance.”

Getting Started: For Staff New to AI

If you’ve never used AI before, this section will help you take your first steps with confidence.

What is AI?

Artificial Intelligence (AI) uses computer systems to perform tasks that normally require human intelligence, such as understanding language, recognizing patterns, or generating new ideas. Generative AI is a type of AI that can create text, images, audio, or video based on patterns it has learned from data.

Key Terms (Beginner-Friendly)

Before you begin, it may be helpful to review some AI-related terms that you may encounter:

  • Artificial Intelligence (AI) – Technology that helps computers do things we usually think of as “human,” like answering questions, writing text, or recognizing images.
  • Bias – When AI’s answers show an unfair preference or leave out important perspectives. This can happen because it learned from patterns in real-world information, which may have gaps or stereotypes.
    Example: If you ask for farming examples and it only mentions large commercial farms, you can re-prompt to include small farms.
  • Confidential Information – Private details about people, programs, or the university that should never be entered into AI tools. This includes names, addresses, financial or medical information, and unpublished research.
  • Context – Extra details you give the AI in the prompt so it better understands your request. The more background you provide, the more useful the answer will be.
    Example: Instead of asking, “Write a gardening tip,” you could add context: “Write a one paragraph gardening tip at the 10th grade reading level for Maine homeowners, focusing on May soil preparation for vegetable gardens.”
  • Disclosure – A short note saying AI helped create something.
    Example: “Created by UMaine Extension staff with generative AI assistance.”
  • Fact-Checking – Double-checking AI’s answers against reliable, trusted sources before you use or share them.
  • Generative AI – A type of AI that can create new content like text, pictures, audio, or video, based on patterns and examples it has learned from.
  • Large Language Model (LLM) – The “brain” inside tools like ChatGPT or Gemini that helps them understand and respond in human-like language.
  • Prompt – The instruction or question you give the AI. Think of it like giving directions to a coworker: the clearer you are, the better the result.
    Example: “Give me three ideas for a summer youth gardening activity in Maine.”
  • Subject Matter Expert (SME) – A person with deep knowledge in a specific area who checks AI’s work to make sure it’s accurate and appropriate.

First Steps

  1. Choose an AI tool
    Start with Google Gemini or ChatGPT. These are the most widely used and supported tools in the University of Maine System. Both work in similar ways: you type a request into a simple text box, similar to a search engine. Features may differ, and both tools are updated regularly, but there is no single “best” option for everyone. For most users, it comes down to personal preference. Skills and techniques you learn with one tool are mostly transferable to the other.
    • Gemini: Every UMS employee and student has free access through their Google Workspace account.
    • ChatGPT: The free version is available by creating an account on the ChatGPT website.
  2. Start with a very simple prompt
    Copy and paste one of the examples below, or try your own:
    • Give me five ideas for healthy snack recipes using Maine-grown produce
    • List ten creative ways to make learning about pollinators fun for kids
    • Create a step-by-step plan for organizing a community cleanup event
    • Explain composting in simple terms for a fourth-grade science class
  3. Review the results
    • You can ask the AI to make changes, including revisions for clarity and audience needs.
    • Add additional context to the prompt. Remember, vague prompts result in vague answers!
    • Check for accuracy, tone, and alignment with Extension’s values.
    • Fact-check claims against trusted sources.
    • If the content is for an external audience, have it reviewed and edited by an SME.

How to Tell AI Exactly What You Want

If you ask an acquaintance, “Pick me up something for lunch,” they might return with a large tuna salad sandwich. While you appreciate the gesture, it is disappointing because they did not know you had already eaten a big breakfast, are allergic to mayonnaise, and eat gluten-free. If you had shared those details and described what you wanted, you would have been much more likely to get a meal you could enjoy.

The same idea applies when working with generative AI. A vague prompt such as “Write an article about gardening” will likely produce a generic result that misses your needs. By adding context such as who the audience is, the level of detail they need, the tone you want, and any specific requirements, you guide the AI toward a response that is accurate, relevant, and useful. Some people call this “prompt engineering” or “context engineering,” but in reality it is simply about giving clear, specific instructions so the AI understands what you want.

Here is a short, helpful video to explain prompting, whether you are using Gemini or ChatGPT:

Tips for Your First AI Session

  • Be specific: The clearer your prompt, the better the output.
  • Try re-phrasing a prompt to see how responses change.
  • Think of AI as a collaborator, not a replacement for your judgment.
  • Always remove confidential information before entering (or uploading) it into an AI tool.

Advanced Tutorials

Monthly Extension AI Roundtable (Zoom)

Extension AI Roundtable is a monthly, drop-in Zoom session for UMaine Extension staff to explore practical, time-saving ways to use generative AI. We’ll kick things off with a short, practical demo or tip, then leave plenty of time for questions, examples, and swapping ideas. No prep is needed, and there’s no pressure to speak up if you’d rather just listen. Come by, see what’s possible, and leave with something you can use right away.

  • Every month on the second Tuesday at 2 p.m. (Look for email reminders with Zoom link)

Additional AI Training and Support

UMaine Extension and the University of Maine System provide resources and training opportunities on the effective and ethical use of AI tools. Employees are encouraged to utilize these resources:

UMS AI Training

  • UMS Discover AI Essentials (Ryan Low, et al. via Zoom)
    Cost: $25-$599
    Master the essentials of generative AI and prompt engineering to enhance education and corporate training. Learn to design course- or training-specific GPTs, integrate advanced GenAI tools, and apply ethical, secure, and effective AI strategies for content creation, assessment, and learner engagement across classrooms and HR programs.

Don’t forget, these programs qualify for staff development funds.

UMS AI Resources

  • Learning with AI (UMaine CITL)
    Provides UMaine faculty with guidance on integrating generative AI into teaching and learning, including syllabus language, pedagogical approaches, citation strategies, and links to additional supports.
  • Learn with AI Toolkit (UMaine CITL and New Media program)
    Encourages embracing generative AI in education by offering a free, crowdsourced toolkit of classroom-ready resources, lesson activities, strategies, and event listings to help students and educators engage with the ethical, pedagogical, and ecological dimensions of AI use.
  • UMaine Research with AI Guidelines (May, 2024)
    Guides researchers, faculty, and students through the benefits and challenges of using AI in research, including automating tasks, uncovering insights, engaging public audiences, and emphasizes responsible and evolving best practices amid shifting policies and ethics.
  • Staff AI Prompting Quick Start Guide (UMS)
    Introduces UMaine staff to practical AI prompt strategies, offering core techniques, use-case examples, and tips that support communication, planning, reporting, and daily academic operations while reinforcing professional judgment.
  • University of Maine System Generative Artificial Intelligence (AI) – Guidance on Use
    Offers guidance on the ethical, responsible, and legally compliant use of generative AI across the University of Maine System, covering privacy (FERPA and HIPAA), intellectual property, bias, accuracy, accessibility, disclosure, and acceptable use, while clarifying that it supplements rather than replaces existing policies.
  • What Uses More?
    Interactive calculator created by UMaine Professor Jon Ippolito, comparing energy and water usage across different digital activities including AI tasks, based on academic and industry sources (with a caveat regarding estimation limitations).

Questions?

If you have questions about these guidelines or the use of generative AI, please contact Matt Thomas, matt.s.thomas@maine.edu.