For audience members who didn’t know the story and the history of Eva Perón — or couldn’t understand every word of the English-language musical inspired by her life — Lawrence High School’s spring production of “Evita” offered a guide.
The show featured images of historical newspaper headlines and articles, which appeared on each side of the stage, one side in English and one in Spanish, as well as archival footage of Perón merged with the face of the teenage actress playing the former first lady of Argentina.
And all of it was created with major help from ChatGPT — so much so that the AI-powered chatbot probably deserved a credit in the program.
“[ChatGPT] was my graphic and set designer, and production assistant,” Lawrence theater teacher Laurie Donlan told me. All of a sudden, she had a partner in her personal ChatGPT subscription. “I was actually doing production meetings in the middle of the night as ideas were coming.” It allowed Donlan, the show’s director, to teach more efficiently, she said, freeing her to spend more time with the kids and nurture their talent. “They won state competitions,” she said.
While school officials around the country debate how students and teachers should or shouldn’t be allowed to use artificial intelligence, Donlan has already put it to work — on her own dime. She’s not alone: A recent national poll of public K-12 teachers, conducted by the Walton Family Foundation and Gallup, found that most use AI but only 18 percent have received formal guidance on using it. (That does not include Donlan, who told me: “Truthfully, I don’t know what our policy is.” A spokesperson said the Lawrence Public Schools is in the process of developing a districtwide AI policy.)
Donlan’s experience encapsulates both the promise of AI in K-12 education and the lack of clear policies that set out what responsible use in the classroom should look like.
That issue brought roughly 100 students, two from each state, to Boston and Cambridge over the weekend for the Youth AI Festival. During the three-day summit — organized by the independent nonprofit Day of AI, the School Superintendents Association, and MIT RAISE — the students discussed, wrote, and approved a model AI-use policy for classrooms, ending in a simulated Senate floor debate and vote at the Edward M. Kennedy Institute.
Among the provisions the students weighed: banning AI use below the eighth grade and for artistic assignments, and prohibiting teachers from entering identifiable student information when using an AI tool to write an individualized education plan.
Their work is ahead of most districts — and most state legislatures. While most states, including Massachusetts, have released nonbinding guidance for the use of AI in schools, fewer than 10 have passed laws actually requiring districts to adopt AI policies. Among them: Tennessee and Oklahoma, which recently passed one of the strictest laws, including banning AI tools from being primarily used for grading, discipline, or other high-stakes decisions, and a provision allowing parents to opt children out of AI tools without academic penalty.
One thing became clear while listening to the students at the festival: While they’re all extremely familiar with AI tools, not all of them are convinced they belong in the classroom. Luci Cox, a rising senior from Marshwood High School in South Berwick, Maine, attended because she wants more clarity on the line between plagiarism and inspiration, a problem she sees in her high school, and to “redefine what ethical use of it looks like.” There’s no AI-use standardized policy in her school, she told me, so the rules typically vary from teacher to teacher. Because of environmental concerns — AI services are voracious consumers of electricity — Cox said, “I don’t use AI for myself.”
Chase Johnson, a rising high school junior from Alaska’s Denali Borough School District, takes the opposite view: AI is “a quite useful tool and we should definitely implement it in our school systems,” he said. Johnson emphasized that he doesn’t use AI tools while in school, only for learning purposes outside the classroom. His roster is long: ChatGPT, Claude, Grok, GammaAI, Granola, Notebook, Manus, Wispr Flow, Gemini, and Base44. “I use primarily Grok for research and Gemini because those tools are better for looking up stuff.” He explained he also uses a mix of tools to make a podcast of a specific subject, which he then listens to while trying to fall asleep.

I asked Johnson if he ever gets concerned about getting wrong information or the AI making blatant mistakes. Not to worry, he said. “A lot of people don’t know but most AIs have a thing that you put in, it’s called a master prompt, and you can put in a set of rules,” Johnson said. “You can make it not make up anything, double-check the sources, and give you the sources.”
Well, then! What Johnson is describing are custom instructions or system prompts, a feature of most generative AI models that allows users to dictate how a chatbot behaves. But, sorry to say, that alone doesn’t make models trustworthy.
Johnson’s comment is indicative of two related but very different concepts that often get muddled: learning or teaching AI literacy vs. using AI to learn or teach.
Consider this: A new research paper that reviews state policies around AI use in elementary and secondary schools found that states are more prepared to procure AI than to teach with or about it, and that most state guidances fail to specify how teachers should integrate AI pedagogically and how teaching with AI differs from teaching students about the technology and how it actually works.
That’s a clear gap that policy makers should pay attention to. Day of AI’s executive director, Jeff Riley, the former Massachusetts Commissioner of Elementary and Secondary Education, told me that’s part of the mission of the nonprofit. It’s “demystifying what AI is,” Riley said. “We have a curriculum built out that shows how to incorporate AI into teachers’ lessons” and “standalone AI literacy lessons.”
In the meantime, the experience of Donlan, the Lawrence theater teacher, offers perhaps the most compelling argument for AI in schools: It can do the work that keeps teachers from teaching. She told me AI technology has expanded her creativity. “AI is already on Broadway,” she said. “And that’s the thing, the stage mirrors life.”
Marcela García is a Globe columnist. She can be reached at marcela.garcia@globe.com. Follow her on X @marcela_elisa and on Instagram @marcela_elisa.
