My daughters were home for the holidays in December of 2022, and one of them asked if I’d heard of ChatGPT?
“Chat what?” I asked.
My response and blank stare led Larkin to pull out her computer. She prompted a site to write an email to an employer letting them know about her plans for a new project. Then she added some details about her project, and the email got better. Then she asked for a more formal tone. Better still. Then she asked for the email to weave in some details and questions personalized for her supervisor…
“Okay,” I said. “How did it do that?”
Since then, Chat GPT has gotten better and better at creating – or should I say producing – written work. It has also welcomed – or not welcomed! – other platforms like it including Bard, Bing, and my current go-to, Perplexity. Simultaneously, AI is emerging with educational platforms such as Brisk Teaching, School AI, and Diffit. When I googled best AI tools for teachers, I had a plethora of options; and there are likely to be more tomorrow.
All of these systems are Large Language Models, and they create content based on the prompt from a user and the vast amount of content and data it has that addresses that prompt.
Over the last couple of years, I’ve given a lot of thought as to how educators can harness the power of AI and make sense of the options, as well as how to teach students to use AI as purposeful prompters and critical creators.
A Few Ways for Educators to Use Large Language Models
Large Language Models have become great work partners for me, and there are many ways I’ve used them in my professional work as a curriculum developer and instructional coach. Here are a few:
To Create Learning Objects and Targets
If you ask the system for X number of learning targets about a specific topic for a specific grade level, it’s likely to produce useful content. If you’re not happy with the content, you can ask it to revise it with specific requests. Maybe you want them in more student-friendly terms. Maybe you want them at a slightly higher level. Maybe you want them more clearly aligned to standards. You type it in, the system produces it. And if you want to revise any of it yourself, the system has no pride of authorship… you can!
To Create Lesson Plans
I’ve been using Perplexity, and when I asked for a lesson plan with these three learning targets, it gives me step-by-step instructions for teaching. It also gives me sources and other websites I could visit that are useful and relevant. I wanted the lesson to include more hands-on activities, and, given that request, it provided some good ideas. I needed the supply list, and… just like that… I had one, complete with where to go to get materials.
To Write Assessments
Prompt engineering is a term I now think about all the time, and it encompasses the idea that the better and more specific you are with what you ask an LLM to produce, the better the product you’ll get. For an assessment, I’ve had useful results with a prompt along these lines: “Write an assessment for a third-grade unit on language standards including 5 multiple choice questions, 3 fill-in-the-blank questions, and 2 short answer questions. Keep the reading level at a third-grade level.”
To Provide Feedback for Written Work
You can copy and paste student work and ask the system for grade-level feedback with a focus on structure or a focus on development. You can ask for a specific number of suggestions, and you can request simpler or more complex concepts based on what you get from the first prompt.
I’m constantly thinking about ways to use LLMs that make my workload lighter, and I’ve learned what I like and what I don’t like about the results I get. I’ve also considered what cognitive work I have to do in terms of creating content and critiquing content. These reflections guide some of my thinking around what students need to know and be able to do as they grow up side by side rapidly evolving technology.
A Few Ways to Use Large Language Models with Students
Teach Students How to Prompt AI to Create an Outline for an Information or Opinion Text
For students who are working on organizing their thoughts, AI can create an outline with a fairly simple prompt. However, when I prompted Perplexity to “Please write an outline for a 5-paragraph essay about why chocolate milk is bad for kids,” the ideas were complex and the vocabulary was high-level. I had to rephrase my prompt to make it more of a 5th-grade level. Students have to be reflective and careful about what they understand and what they don’t for this possibility to be of any benefit.
Support Students in Using AI to Find Research and Information on a Topic
AI has been an important work partner for me for finding studies and information about topics, but there’s a lot of navigating and curating I’ve still had to do. More than ever, it’s important to know about reliability and the differences between information sources. While Perplexity cites its sources, students should know the different uses for magazine articles, brochures, encyclopedia entries, and peer-reviewed research.
Help Students Use AI to Check Written Work for Spelling and Grammatical Errors
Many students have already been using AI for this purpose; grammar and spelling checks have been around for a while. It’s important to know their limitations since sometimes voice or intended meaning is lost in the suggested changes.
AI and Large Language Models have definitely opened up new dimensions within education, offering powerful tools for both educators and students. More than ever, balance and reflection are important. How do we leverage AI’s capabilities to improve educational processes, and also maintain the human elements of teaching and learning? Preparing students for the world and their lives beyond classrooms has always been a beacon for educators, and now, we’re mixing more elements into the complexity of the future.
Join us to learn more from Melanie at our next Reaching All Learners Speaker Series session on Wednesday, March 12, 2025. Register here!
