Need an artificial intelligence policy for your schools? This district used ChatGPT to develop it

Last July, the Peninsula School District in Washington state became one of the first school systems in the country to release guidelines for using artificial intelligence in the classroom.

Since then, the district's website has become a busy corner of the Internet, with its "Principles and Beliefs of Artificial Intelligence" garnering more than 600 views per month. This is likely due to the fact that the document has been shared by nonprofit organizations such as TeachAI and the School Network Consortium seeking to showcase AI governance models to other districts.

Chris Hagel, executive director of digital learning for Peninsula County, who participates in webinars and forums on the use of AI in K-12 schools, makes no secret of the fact that he had an unusual co-author on an AI guide: ChatGPT.

The AI tool helped develop the first draft of Peninsula using data provided by Hagel. Hagel and his team then edited and improved the document, a writing process that Hagel believes could become more common as generative AI is implemented.

Pennsylvania, which released its guidelines last July, is unusual. About 80 percent of educators say their districts have not developed clear rules for using artificial intelligence in the classroom, according to a survey of 924 educators conducted by the EdWeek Research Center in November and December.

Hagel believes school districts need to learn how to use the technology themselves - and get their students and teachers to use it, too. The Pennsylvania guidance notes that AI can help educators with routine tasks and meet individual student learning needs by helping present information in a variety of formats. It also recognizes AI's shortcomings, including algorithmic bias, and states that humans should have ultimate authority over any decisions made using AI tools. In addition, the paper recommends that students should be trained in the proper use of AI.

We spoke with Hagel at Zoom about setting expectations for the use of artificial intelligence.

Why did your county issue an official AI policy so far ahead of everyone else?

Teachers [using artificial intelligence with students] asked for this so they wouldn't be the only ones sticking their necks out. They said, "The district needs to tell us something so that I have protection if a parent complains, or a student doesn't want to do it, or my principal evaluates me and they don't like what I'm doing."

It is easier for teachers to move forward if the district says this is what we believe and what we accept.

How did you develop the artificial intelligence guide?

Just as teachers were seeking advice, the U.S. Department of Education released a document on teaching and learning with artificial intelligence. This was one of the documents I used as the basis for our guide.

I highlighted in a 75-page PDF document [ED] everything that I thought was important. Then I took all of those highlights, as well as a couple other documents that were released at the time on generative AI in education, threw them all into ChatGPT, and said: "Make me a draft of the principles and beliefs document."

I then took four pieces of my own writing and asked ChatGPT to analyze them and then rewrite them in my voice and tone.

This was the first draft. We then gave it to the Teaching and Learning staff to look at and give feedback. Then I gave it to our teachers who were working with AI and asked for their feedback. And at that point we just [released] it, saying: "Here's what we're thinking of doing. If you're interested, let us know."

What advice do you have for other districts that want to develop their own AI guidelines?

The first thing I tell people is to just steal ours. [Laughs]

Honestly, I tell them not to spend a lot of time thinking about it. There are tons of examples. You can take ours. You can take a guide from TeachAI. Already six states have put out a guide.

One of the challenges facing AI right now: you don't know what's coming next year, you don't know what's coming next week. Every day, any major tech company can release some new amazing AI development. And then you think, "Oh, wow, that picks up on everything we've already talked about."

So I advise people to just get started. Historically in education, everybody wants to set policy, develop guidelines, and then get into professional development, implementation, and buying programs. And we do it kind of in reverse order.

Let's give it a shot. And then we'll figure out what works. And then we'll write rules for it. And a lot of people are very hesitant to do that.

There are a lot of superintendents I've talked to who say, "I need standards, I need guidelines, I need the state to tell me what to do. And then I'm going to do it in my district."

I think the problem with AI is that you can fall far behind. You'll have a whole class of 12th graders who never learned anything about AI. They will leave your system. As you continue to wait because you're not ready, kids who you could have done something about will leave your system.

At what stage are you in the process of implementing AI in the classroom?

Have you seen an innovation curve where there are early adopters, then fast adopters, then everyone else? And at the end, the laggards? Right now we are in the second stage [fast followers].

Our early adopters have embraced AI. And now we're trying to figure out how do we get the bulk of teachers to embrace it? We now have about 10% of staff who are quite comfortable using AI in the classroom.

I just met with a group of our teachers yesterday and they are expanding and developing what they are doing. They are starting to change the way they teach in the classroom. A high school chemistry and astronomy teacher was talking about how he expects every student to use AI. And he is suspicious that when they submit work, they don't include a link to their ChatGPT conversations [for reference].

We embrace AI as a district. Our board of directors supports our moving forward with AI. We've identified the core principles of working with AI. Essentially, we have the right framework, and now we're trying to figure out how to expand and evolve it.