Policy Guidance for Generative AI just Released

Cover of Guidance Document

From the time I started with my interviews to the time I got through the review, the number of states that had released some sort of AI guidance grew from just a couple to a little over a handful, with two states releasing guidance as I was writing! I reviewed some regional and national guidance documents, as well.

My goal was not to write the definitive “how to” guide on developing AI guidance or policies. I’m not sure anyone can do that at this point, because AI is evolving so quickly. Instead, I tried to learn from those states who first tackled this important issue in order to offer guiding questions leadership teams might consider as they develop their own guidance and policies. For this reason, I call them considerations. The considerations are written in a way that would be suitable for leadership teams at the school, district, or state level.

Not every school, district, or state is going to need to address every consideration. Based on the documents I reviewed, not all states did. But taken as a whole, we can learn from across all of these states—and those that have and will release guidance since this was published—to strengthen new policy documents as they are developed. It was interesting to note the influence of early documents on subsequent documents released afterwards. That’s the spirit of this document: learn from others and move our understanding forward.

I hope the considerations are helpful. They will eventually need review and revision as the capacity and characteristics of AI change over time. Some states astutely labeled their guidance as “1.0.” That’s a good lesson for anyone delving into this realm.

Building My HI about AI: Part 3 with John Daniels

Helping Teachers Explore AI at the Classroom Level

This series opened with a conversation with my friend and colleague, Eric Nentrup, who is an educational consultant with his hand on the pulse of AI (artificial intelligence) in education at the national level since he has and is continuing to help publish reports on AI for the Office of Educational Technology in the U.S. Department of Education. I then focused my attention on school districts by chatting with another friend and colleague, Maria Stavropoulus, who has leveraged her statewide connections and is championing AI at the district level. She’s even teaching a college class on using AI in education. I thought it would be nice to round out this series with a discussion with someone working directly with teachers and students in a school, and I knew exactly who that should be. My goal with these conversations is to increase my HI (human intelligence) about AI—and create a path for others in education to follow.

This past school year I’ve had the good fortune to work with the instructional technology facilitators (ITFs) in the Carteret County Public School System in North Carolina. The ITFs and I have been working on our coaching conversation skills and practicing coaching rounds with teachers in their respective schools. In talking with several of the ITFs, it became apparent that many of them are interested in AI and some are taking steps to help their teachers use AI in their practice. One of those, John Daniels, is the ITF for Newport Elementary School and the perfect person to talk with about AI in the classroom. When we talked about it last fall, his interest and excitement was palpable. Little did I know I would have the opportunity to learn more about AI from him with this blog series.

Getting Started with AI

John told me he has been using AI personally for several years, and has had some success helping teachers in his building explore using it in their classrooms. He first heard about AI through the state’s technology conference, North Carolina Technology in Society (NC TIES), during a presentation on ChatGPT. He had tried ChatGPT a little before then, but the presenter made it sound “really cool and useful,” so his use of that and other AI tools started in earnest then. He notes that he now uses some type of generative AI (GenAI) pretty much every day and is finding more and more ways to help teachers use it. He finds it “very, very useful.”

I had to ask what he does with AI every day. I just couldn’t imagine using it every day. John had several examples. He notes that he does some writing most days and he uses AI to “organize his thoughts and get things the way I want it down on paper.” One of the things he writes most days is a script for his elementary news broadcast team. John said AI saves him a lot of time generating those scripts. He described an instance where he needed a blurb about a character trait for the students to share on the broadcast. It was an opportunity to support academic and social outcomes for students in the school. Using AI, he was able to synthesize some examples and add them to the script quickly.

Coaching image generated by AI 3/14/24

Image generated by AI 3/14/24

He’s thrilled about the opportunity to help teachers use AI to individualize instruction for the various needs of different students. This echoes both of my other conversations, which shared the potential of AI to truly personalize learning. John noted that when he or other teachers work with a student, they find out what their interests are “and then you can very easily take a lesson that you’re working on and tailor it to that student, in really just a few seconds.” He said you can do that for individual students or groups of students in a short amount of time to “make the learning special and appropriate to them.”

He’s also had success helping teachers re-level reading passages so that different students can access the same content but at a developmentally appropriate reading level. I told him my struggles with trying to get reading passages I wrote to a kindergarten and first-grade level using an older version of AI. He understood the issue, saying that he finds it worth paying for the newest versions of his favorite AI resources. He notes that there are some free AI sites online, but many of them offer additional functionality and resources at the premium level.

Helping Teachers Use AI

Much of John’s work is with teachers, because at an elementary school many children are too young to access some of these resources—at least at the current stage of AI development. That means he has a lot of great examples about getting teachers to explore AI. Most teachers have to create or find a lot of lesson plans throughout the year. When I was a teacher, every Sunday afternoon was lesson plan time and I would spend hours getting ready for the week. Now, John said, teachers can use an AI resource like Magic School and not only cut down on the time it takes to find or create lessons but also customize their lessons and make them more meaningful to their specific students.

I had previously talked with the ITFs in Carteret County about the importance of providing learning opportunities for students that get to the point of learning transfer, what some might call deeper learning or strategic and extended thinking. John took this to heart. He notes that teachers can feed lesson ideas into a resource like Magic School and it will offer suggestions and make connections that push the cognitive complexity of a lesson. He explained that “you’re not going to use the material verbatim from what it generates,” but it helps teachers get further along in their lesson planning faster than starting from scratch.

Lesson planning help sounds like a great time saver to me. I think about all those Sunday afternoons I spent on it, but John said there’s more. The AI resource Diffit also helps teachers find or generate lessons but then formats them so they can easily create content that it then imports into Google Classroom, which is the learning management system (LMS) his teachers use. It can also generate an interactive Google Slides deck on a topic of study that teachers can then customize and use in their instruction. This harkens back to the “automate the mundane” quote that Eric Nentrup shared. It’s a great example of saving teachers time on something they know how to do and need to have done, but don’t need to do themselves. John said that for every teacher he’s shown this to, they call it “a game changer.”

Having lesson activities individualized and imported into an LMS has then increased the district’s return on investment for the hardware purchases they’ve made. The district provides Chromebooks to students, and John said that when it’s so easy to generate lesson resources and make them available to the students, teachers and students end up using the Chromebooks more often. He notes that once he has teachers and students comfortable using the Chromebooks, he introduces ways to leverage them even more and to add additional resources, whether AI or not. Talk about great coaching. John has helped some of his teachers address a fundamental need that then provides an opportunity to continually grow their skills. So, AI is not only impacting teaching and learning, but John’s coaching and professional learning opportunities in his school.

The skill of prompt engineering came up in my conversation with John and is one of those “even more” skills that he helps teachers work on after they feel comfortable with some of the foundational tools like Diffit. All three of my experts agreed that prompt engineering is an important skill for making the most of AI. After Diffit, John suggests teachers explore Magic School AI because it provides support for better prompts. “It puts the specialized prompts together for you, and that’s why it gets good results.”

Once a teacher feels comfortable at that level, many move on to generating their own more effective prompts and can use other tools that expand the use of AI in other areas, whether that’s customizing language for written feedback, crafting emails more quickly, contributing to a student’s Individualized Education Program, or even generating new images they can use in presentations or handouts. Many teachers can use their new prompt-engineering skills with other tools like ChatGPT or the new chatbot Perplexity.ai, which don’t provide templates or support for prompts.

Lessons from Teachers

It’s obvious that John is enthusiastic about AI and its potential, and his enthusiasm is helping others find the confidence to try some of those tools in their classroom. I asked if any of the teachers had any “aha moments” when trying out an AI resource. Coincidentally, I was actually in the classroom with a teacher during our coaching rounds who John said had indeed had an aha moment about AI. Before the visit, John was showing some resources to some of his teachers when one realized, “Oh, I can use this for Cloze questions.” Teachers use Cloze Reading Passages to help students determine or confirm the meaning of words using context clues. The teacher was already using this practice with students, displaying Cloze passages to the class using a projector—and a light bulb just went off and he realized he could use AI to generate new ones or quickly customize them to what they were reading.

John acknowledged that on the surface level, using AI in that way may have seemed like just a simple thing, but for that teacher, that was the point where he said, “Okay, I’m going to learn more about this.” That was that teacher’s first step toward using AI—and I got to see it! But for me, I didn’t notice anything except a strong teaching strategy. The AI part was transparent. But that first step has also opened the door to more coaching conversations between the teacher and John. Wins all around.

From his own experiences, John said educators really want to understand the benefit of a new resource before they invest too much time or effort in it. Is it something that’s just going to be replaced in a couple of years? John said that the benefit teachers note almost immediately with AI is the time-saving aspect of using these tools to support their instruction. But he said that’s a stepping stone. Yes, teachers save time replicating things they are familiar with, but John kept mentioning the number of teachers who then see how they can better personalize their instruction. It gives teachers “more of a chance for all of their kids to be successful.”

On a more pragmatic note, John shared that it’s important for teachers to understand that they must never share any personally identifiable information (PII)—their own or their students’—when using GenAI resources. There’s no assumption of privacy when using any of these services or models. He cautions that teachers need to be sure they aren’t sharing any student information when they are using AI to generate feedback. He notes that there are some companies that are creating their own models that may be able to help keep PII secure. He mused that even the state might be able to generate its own model that’s locked down for privacy and provide a safe environment for all teachers and students in the state to benefit from AI.

Support from the District and State

As I mentioned, many of John’s colleagues in Carteret County are interested in leveraging AI with their teachers. The ITFs get together as a professional learning community at least once a month for an opportunity to collaborate and share ideas. At their most recent meeting, Mike McKay, the district’s chief technology officer, spent time with the team exploring AI from the district’s perspective. As a district, they are definitely aware of AI and developing their own knowledge and skills around its potential. I enjoyed hearing how IT is not the only department having “the AI conversation”—to quote my colleague Maria—but is involving those directly responsible for teaching and learning.

John said that he’s found a lot of support for AI at the state level. Not only has it been featured at the NC TIES conference, the state has taken a proactive stance and developed North Carolina Generative AI Implementation Recommendations and Considerations for PK-13 Public Schools, putting it ahead of most other states in the country when it comes to AI policy. He has found the work of Vera Cubero and her fellow digital teaching and learning consultants at the state’s Department of Public Instruction to be especially helpful. Vera is helping others learn about using AI by modeling AI in her own work, such as this comprehensive and helpful presentation.

John gives credit to Vera for the CRAFT prompting protocol that is gaining widespread adoption within and beyond the state to help staff become better at the important skill of prompt engineering. CRAFT stands for Context, Role, Audience, Format, Task & Tone. There’s more about CRAFT and other AI issues in her presentation, like why AI detectors are problematic and how to create AI-resistant assignments.

Coming full circle, John is paying it back to that AI presenter he saw at the NC TIES conference last year by leading his own conference session on AI at this year’s conference. It’s called Use of Artificial Intelligence to Personalize Student Instruction. Based on the examples he shared with me, it should be a terrific session that may start other educators on their own AI journeys.

What Do You Hope for the Future of AI?

As an enthusiastic user, I asked John what he sees for the future of AI, whether for him personally or for the teachers in his school and district. He thought about it for a while before responding. At the forefront of his thinking is that everything AI-related is moving so quickly. He noted that depending on how you approach it, that can be a good thing or a bad thing. Job loss is a concern. AI is already making a lot of jobs redundant, but then again it’s generating new jobs and career opportunities.

He said that AI detection tools were popular for a while but that many universities have stopped using them or are turning them off. Echoing Eric Nentrup, John noted the initial knee-jerk reaction of focusing AI on plagiarism wasn’t a productive strategy and even led to some lawsuits where original work was misclassified as AI-generated. He compared the introduction of AI to the early use of calculators in schools. He remembers that when calculators were becoming used more often in schools, people had concerns about students using them to cheat and wanted to ban them from the classroom. But now there are calculators everywhere—on your phone and on many digital devices. “There are times when you use them and there are times when you just don’t use them, just like there are times when you don’t use AI. You [educators] have to set the standards and help students know when they can and should not use AI,” he said. As with his personal use, he said it may be appropriate for students to use AI on some assignments to organize their ideas or even generate new ideas, but not on others. “We’ve just got to teach kids how to do it effectively.”

Teaching kids how to use AI effectively can include lessons in which kids share how they have used AI in their response. John says teachers can ask, “What did the student use AI for? How did the student take what the AI resource gave and modify it to fit their perspective? How did the student go back and edit their prompts?” He sees students going beyond just getting a response from a prompt. He wants students to be able to think about and share how they used AI effectively to make their work better than what they could have done on their own. Remember his earlier comment about cognitive complexity? When a student can reflect on their own work, critique the resources they have used, and generate strategies for improving their work in the future, they’ve moved beyond simple recall and skill demonstration to strategic and extended thinking that is the hallmark of learning transfer. Well done, John Daniels.

This blog series has been a great push for my own AI journey. I know that there are organizations and experts that I trust who are providing guidance for AI policy and practice to states, districts, and teachers. I also learned about—and even played with—some new AI resources to get a better feel for the range of tools available to support educators. It’s also affirming to hear the varied experiences of three people I know and respect, and see the overlap in how they can help me and others better understand and leverage GenAI. I’m certainly still not an expert, but hearing how these three have increased their expertise shows me that all of us have opportunities to continue to learn and benefit from AI. I’m sure I will continue to learn from others, as well.

Want to know more about John Daniels? John’s days are full working with teachers and students on a variety of topics, not just AI. He also specializes in STEM, supports the student news team, and is his school’s go-to person for anything with technology. You might have missed his AI presentation at NC TIES this year, but be on the lookout for more shares from this dedicated coach and teacher.

Build Your Own HI about AI by Exploring these Resources John Daniels Mentions

Please note: not all resources are free

North Carolina Generative AI Implementation Recommendations and Considerations for PK-13 Public Schools and the accompanying presentation Deep Dive: NC Generative AI Guidelines

Magic School AI provides templates for creating lesson artifacts

Diffit generate lessons in interactive formats

chatbot Perplexity.ai