Skip the Prompts and Just Talk to AI

Vicki Davis (00:00): Are you still trying to figure out the perfect prompt? Today’s guests say prompt engineering is a dated thing to teach, and you’ll be relieved to hear why. Have you ever wished you could head to lunch and come back and find some work done? Today we’re talking about ChatGPT for teachers and the AI tools beyond it in plain friendly terms. This is your Tech Tool Tuesday, episode 1017.

Announcer (00:29): This is the 10 Minute Teacher Podcast with your host, Vicki Davis.

Vicki Davis (00:33): Today we’re talking with two people who think the thing we’ve all been teaching teachers to do with AI is already out of date. Christie Cloud is a founding member of the EduGuardians, a community of educators that has grown to more than 240 members.

Dave Hinrichs is director of digital platforms and pedagogy for Johnston County Public Schools in North Carolina, where he has spent 30 years in one system. Together they founded CraveEdventures and they’ve just contributed a chapter to The Educators’ AI Guide 2027. So Christie and Dave, welcome to the show. I want you to respond to the schools that are banning AI in the classroom.

Should schools ban AI in the classroom?

Christie Cloud (01:19): For the schools who are currently banning AI, I understand that AI can be problematic when we put our students in front of it. And I understand that we need the utmost care when dealing with our students. But the entire purpose of education is to prepare students for the world that they will one day inhabit as adults. And there are very few jobs now that are not including AI in some form or fashion. You can’t do a Google search without AI being included in that.

So while I understand the motivations for our students, I believe that it’s rather short sighted because in order to prepare them for what comes next in their next chapters, we really need to be teaching fluency and safety, and they need to have their hands on it. And use it in order to understand all of those factors.

Vicki Davis (02:21): Okay, Dave?

Dave Hinrichs (02:23): To answer the question about schools banning AI, I think like most things, a nuance. When you take an extreme on either, you don’t get to what really needs to be there. And as someone from a district level, writes and updates the guidelines. Yes, there’s gonna be concerns about PII and data privacy and safety on one end and from a classroom level, there’s gonna be the worry about cognitive offload, but it can’t be a outright ban and of course we can’t have wild, wild west.

So there needs to be really clear instructions. And I wish I knew who I heard this originally from.

But it resonated with me and I say it a lot. We messed up. Cell phones. We said, it’s not our thing. That’s a home thing, but they’ve impacted us. We’ve messed up social media. So it’s not our problem. Home, but it’s impacted. It’s all become integrated, connected to people’s lives. This technology is the most transformative since the internet itself and we can’t mess it up. So we really, and it’s moving so fast.

And that’s, you know, that’s how we started the chapter. We blinked and AI changed and it’ll change again. While people are still learning to prompt, we’re over here with agentic systems. We have to be able to stay on top of it, and make sure the students are still learning.

Vicki Davis (03:42): We’re talking about AI as if it started November 30, 2022. We all know that it’s far before that and all of us who’ve been in it a while have been talking about AI for a very long time.

But we’re specifically talking about LLMs for just a minute. You know, when we started and everybody was all so, gung ho on teaching prompt engineering, you know, for me I’ve never been gung ho. It’s like it writes prompts for you. It’s a word calculator? Like it can help us. And of course, the interview method of having it interview you is one of the best methods. Let’s get to the argument of your chapter of like, hey, the stuff we’ve been teaching about AI is not really what we need to be teaching about AI.

Do teachers need to learn prompt engineering to use ChatGPT?

Christie Cloud (04:23): I think Vicki you summed up some of the points that we are making so pointedly is that so often it’s, click here to get 50 prompts for teachers or download this guide for and it becomes a conversation the more you use your LLM of choice. It learns significant factors about the way you speak and what you need, it becomes more of a conversation.

And we believe that a natural tone in conversation with your LLM of choice is going to lead to better results. And also it’s gonna be a lot less scary for teachers who are beginning the process. When you tell someone, you need to use these verbs or you need to structure with a role and suddenly you start losing people and they sort of get like a glassy gaze and they’re like, I don’t I just don’t know what to do. This is beyond me. And that’s kind of what we want to really work to eliminate.

It’s not beyond you. No matter what stage of acceptance and integration with AI you are, like you can do this. It’s really just learning to chat with it.

How do I find my starting point with AI as a teacher?

Dave Hinrichs (05:38): When we set out to write the chapter, one of the things, it sort of stemmed from lots of conversations. One of the things that Christie is really big on is talking directly to teachers. It’s important to her. We’ve been at conferences standing in lines and she’ll pull up teachers’ eyes say, what do you want to see? What’s important? Been trying to interview as many people as we will just for the sake of that, learning and building.

And one of the things we noticed is there was just a wide level of understanding of AI. In front of a group, I’ll do fist to five, your understanding. And so many times people will say one and then you start teaching and they’ll be like, you’re really good. You’re way more than a one. We didn’t have common language.

And then somewhere like five and I start showing things and they’ve never seen it. Go back to what Christie was kind of saying, the conversation, the word prompt engineering was scary. Wasn’t really engineering, it was just talking, like you said, calculated for language. Often say you don’t need these big frameworks or to have it intimidate you. Don’t know how to prompt it and give me these prompts.

Just ask you two or three questions and say, what don’t you understand? And this is my unlock. Ask me one at a time so you can steer it’s incorrect and also don’t get a wall of text. But just helping people understand that there’s different ways and levels of using it. And what’s not better than the other? There is more than prompt, get back, prompt, get back. We wanted to help guide people understanding to more complex uses. I tried not to say advance. Because sometimes you just want a question answered and sometimes you’re orchestrating a system, but more complex.

And when I set out to learn it originally there wasn’t a clear pathway. There was a YouTube video here and an article here and none of it was coherent. And so we tried to lay out, hey, find where you are and here’s how you can move along.

Vicki Davis (07:25): And so let’s go to like this moment. How are more advanced people perhaps using AI. Dave,

How can two teachers share one AI system?

Dave Hinrichs (07:33): Sure. So.

Personally, our model of choice is GPT. And has shared context for all manner of things. But sometimes we want to get her AI’s opinion on my AIs and back and forth.

For example, we have a number of conference sessions that we’ll submit. And we have those stored in there as canon. And then if we want to adapt them, we’ll say, I need you to print this out for, let’s say, FETC and NCTIES. Well, they’re totally different formats. And it has those formats saved so it can spit them out. And then we’ll build them together, iterate between different AI tools.

So what we decided to do was try to build the multiplayer AI system where we have our personal context separate, and our shared one where it can interact with all of our open loops, our projects, things we’re trying to create, social media strategy, conference, proposal sessions, and we can both work out of that together to iterate.

Can ChatGPT’s Codex app do more than write code?

Dave Hinrichs (08:35): And so it’s a really exciting unlock that we’ve had to play with. And we do that through Codex rather than going to just ChatGPT. The Codex desktop app will let us manipulate the files.

It’s why I moved away from a custom GPT long ago and lo and behold, now they’re about to deprecate them because as I was iterating, I get a new idea and get excited about it. And then it would be gone unless I went in and updated all the documentation. So it was just, I didn’t want to use it anymore as a source of frustration.

So I started working at a Codex, not just to code, but to just build. Because I like to say, it’s AI. Instead of a brain and a mouth, now it has hands. And it can do the things for me that we come up with. And so we just made that our default. And it helps us build. And so that’s what I’m really excited about, that and idea of building anything you want. We call that designing the impossible.

We have an origin story about that. We had a tour for a conference session that was the metaphor of the whole session. And then the night before, we were locked out of this little 3D tour. So Christie heroically built another one in a different tour. It wasn’t quite as good. But it was still amazing because that’s all she does is amazing. It makes this in tour. She’s like, I’m to have to use this and really serve the function.

The next morning, I woke up and said, I wonder if we can build this. And so right after the conference got home, we started building it. Now we had our own 3D tour maker. And then I was like, I think we can build just about anything, which is what we asked now. Can we build this? And then should we build this? Because sometimes we don’t need to. There’s something better. So that just, solving whatever problem on the fly and the multiplayer AI. And you said what you’re excited about. I’m very excited about both. That’s my geek out moment.

Do teachers need the most powerful AI model?

Christie Cloud (10:23): One of the most important things to remember when using AI is that you don’t always need a Ferrari. I am very excited about using agentic workflows and to combine our context, but on a day-to-day level, that’s not what I use AI the most for. And in our framework, we try to put forth that message that you don’t always need to use the top model to get the output that you need. And you don’t always need to be in an orchestrate mode in order to get what you need.

I use AI a lot in my role as an ITF to try to help my teachers solve problems that they’re dealing with in the classroom on a day-to-day basis.

How can a teacher use ChatGPT to build an escape room lesson?

Christie Cloud (11:10): We have Constitution Day coming up, and the teachers were like, What? You know, this isn’t really in our standards, so how can we integrate this? And I was like, you know what? Let’s see what we can do. So we took some really good items that we had that really explained and gave some context to AI and I used ChatGPT Work. I wasn’t even in Codex and made a really great escape room that the students could do interactively in class. I built it with, a teacher’s guide and it had intro slides to introduce the activity.

And that’s another thing that I love about iterating with AI is once you have the one thing that you need, sort of the core that you’re building your lesson around. We all know from our experience in the classroom, it’s just part of it. You still have other things that you need. And by working with AI, you can create those other things at a speed that it was previously just impossible to imagine.

So now it’s like, you want a one pager to go with that? Yes. Let’s make sure that we feed it some really good context. Let’s make sure that we check its output for accuracy.

How do teachers keep the human in the loop with AI?

Christie Cloud (12:22): But it’s gonna do a lot of the hard work that we spent so much time putting in. And then we’re there, because the humans always in the loop. So we’re there at the beginning to make sure it gets the best information that it needs. And we’re there at the end to double check it and make sure it’s giving us exactly what we need to meet our classroom goals.

And that’s what excites me the most about it is that all of those things that were so time consuming for teachers I can do this in a few clicks. And then my expertise that I’ve spent years honing and a great education, getting to this place can be used for what I really want it to be used for, not time consuming putting together piece, but the overarching, like curating piece. This is what I need in my classroom.

Vicki Davis (13:12): How do you simply explain AI agents to teachers who are like, okay, so I don’t even understand how to use a regular AI. Now you’re talking agents?

What is an AI agent, in simple terms?

Christie Cloud (13:23): Agentic workflow is when you are directing the AI and the AI is doing the work for you. It’s often very interconnected and it is moving toward a goal that you have set for it.

Dave Hinrichs (13:37): We often talk about student agency, right? And we get what students to be able to have some modicum of control over their learning. But we don’t say, whatever you want. We still give it the context guardrails. Now the AI, rather than just having a conversation, now has some agency to go achieve that goal as well. But yet you’re still giving it the context or guardrails.

Vicki Davis (14:00): For me, when it’s an agent, it’s pretty much something like, hey, I’m gonna give this to you and walk away for an hour. It will text me on my phone and say, I do need approval for this. It can read my email, it can draft email, it can’t send email. Give us some other tasks that you have your agentic AI do.

What tasks can teachers give an AI agent?

Christie Cloud (14:19): One of the things that I have instituted because I like to do weekly tech tips. I have it collect and then provide to me when I ask for it some updates, usually some Google things that are happening out there. And then it pretty much summarizes that for me, gives me some links that I can check out and helps me curate my email to send out for that. It’s been super helpful. So that again, I’m saving a lot of that time that I would be checking all of these different websites to get that.

Email is also one that I have very much enjoyed using Agentic Workflow for because I can get a summary, while I’ve been gone, what’s coming, what needs my attention the most. And that’s been really helpful because I think the one thing we could all agree on as teachers is there’s never enough time.

Vicki Davis (15:12): So thank you, Christie and Dave, for having that conversation with me today.

Christie Cloud (15:15): Thank you so much for having us.

Dave Hinrichs (15:17): Thank you for having us.

Announcer (15:18): Thank you for tuning in to 10 Minute Teacher Podcast. Join us here every weekday and subscribe to the Classroom Matters newsletter. See you later, educator.