Educators around the world are already using ChatGPT and other generative AI tools.

Jennifer Chen, EdD, a leading early childhood education researcher at Kean University, is developing a framework to guide purposeful and ethical implementation of AI in educational settings.

“Rather than replacing their work, AI assists with routine, administrative, and instructional tasks, giving educators more time to invest in what they see as the heart of teaching: cultivating caring relationships, providing individualized support, and engaging in the uniquely human dimensions of teaching and learning that technology cannot replace,” Chen says.

But overreliance on these tools to develop a curriculum and lesson plan, Burns worries, could lead to alienation from one’s work. “When I was designing activities for students, you know, I’m actually thinking about my students,” says Burns. She thinks about the strengths of individual students and whether assignments or projects would suit them. “To me, the lesson instructional design is the most creative part of teaching.”

Chen also cautions that AI-generated plans may contain factual errors, historical biases, and other perspectives that may need to be reviewed or revised by educators.

“AI can be a powerful tool for supporting teaching, but it should augment educators’ expertise—not replace their professional judgment, creativity, or relationships with students,” says Chen.

While experts have suggested that AI could enhance learning through personalization and novel teaching methods, they’ve also raised concerns that it may pose risks to children’s cognitive development, creativity, mastery of school subjects, and social and emotional development. However, they also raise the urgent need for more high-quality research on the effects of AI in children’s education and development.

Burns worries that students will use AI chatbots to cognitively offload the mental effort of completing assignments and miss out on learning. “You have to know how to learn to do basic skills that, as an adult, we might eschew and say, ‘Well, that’s ridiculous. That’s a waste of time.’”

In her interviews, some kids would say they don’t want to read a book when AI can summarize it for them. “What we’re seeing in schools is the kids are offloading their skills and their knowledge before they’ve even onloaded it.”

Many students are also using chatbots as companions, though some experts worry that chatbots prioritize engagement over well-being. Burns said that while previous media developed for kids by the Public Broadcasting Service sought insights from experts and psychologists to develop appropriate programming, that isn’t the case for these generative AI tools, which are frequently marketed to young learners.

Sara Baldassar, MEd, a Cleveland, Ohio, grade 6-8 science and social studies teacher, is heading into her 28th year of teaching. She says that students sometimes don’t realize they’re interacting with AI. They don’t know, for example, that the results at the top of a Google search are AI-generated.

“I’m really, really adamant about making sure that my kids recognize when AI is being used because that’s part of being a good digital citizen,” she says. “Being able to go, ‘hey, wait, this is not true, this is not real, this is AI-generated.’”

Although educators and experts have also raised concerns that these tools will fuel more cheating, a 2024 survey suggested that cheating behaviors may have remained fairly stable among high schoolers since ChatGPT’s release. “I don’t know that I’ve necessarily seen an increase in that kind of behavior,” says Baldassar. “Definitely, they’re using it more than I am comfortable without having any kind of digital literacy training.”

Because of their broad-ranging impact across many employment sectors, Baldassar believes that students will need to learn to use these tools.

Generative AI rolled out before teachers and school districts had guidelines and training in place, which may have led to educators unknowingly inputting sensitive information into public models.

As a result, some school districts, like Baldassar’s, moved to completely ban AI. New legislation in Ohio requires school districts to have an AI policy in place by July 1, 2026. Baldassar, who sits on the Cleveland Teachers Union, helped develop a policy for her district that would train staff and work toward building digital fluency in students.

The American Federation of Teachers (AFT) created the National Academy for AI Instruction in 2025 to help teachers learn about AI so that they understand the risks and potential benefits. Rob Weil, MA, CEO of the academy and director at the AFT stated over email that the academy has already trained 12,000 educators, with a focus on preparing them to use AI safely, responsibly, and effectively. “Educators value learning not only how to use AI, but when not to use it,” says Weil.

The academy has received some criticism for the US $23 million in funding it received from Anthropic, OpenAI, and Microsoft. “The funding, I suspect, is for these companies to make further inroads into the education market,” says Burns. “However, having a union lead the AI trainings engenders a lot of trust—though for many educators, it may signal capitulation or being co-opted by big tech.”

Weil says that they value their relationships with these companies but emphasizes that the academy is independent and educator-led.

“We will work with independent researchers to study broader impacts on teaching and learning,” says Weil. For them, successful implementation means ensuring educators make better decisions about AI, reduce their workload, and protect students, while also ensuring that the tech strengthens rather than replaces teaching.

Burns is concerned that these courses approach AI through an adoption framework, leaving little room for those who want to abstain due to legitimate concerns about the technology.

Implementing these tools as quickly as possible also doesn’t make sense to her, especially if AI is here to stay. The framing of “panic and falling behind that’s pushed onto these education systems, I don’t think lends itself to kind of thoughtful and meaningful and purposeful integration.”