Part 1 of TED Radio Hour episode "Are the kids alright? Part 1"
As a new school year begins, teachers and students find themselves at a perplexing crossroads: how to use artificial intelligence in the classroom.
A 2024 study from the Digital Education Council found that 86% of students globally use AI for schoolwork, with 66% utilizing ChatGPT regularly. And a recent Gallup poll found that 3 in 10 teachers also use AI weekly to prepare lesson plans and grade assignments.
These figures raise a number of questions: Is using AI considered cheating? How much is it supporting or hindering learning in the classroom? How will it affect kids' analytical skills? And how else can AI be used in the classroom in the coming months and years?
Technologist Victor Riparbelli predicts a future where students interact with AI avatars rather than read and write. "I think we're at the dawn of a new era of AI-enabled communication," Riparbelli said in his 2024 TED Talk. "And one day, I think we'll look back at reading and writing as historical artifacts, like we do with papyrus scrolls or hieroglyphs or cave paintings."
Less reading and writing, more AI avatars
Riparbelli is the co-founder and CEO of Synthesia, a company that uses AI to create video lessons led by lifelike avatars. The company was valued at more than $2 billion in January 2025. These avatars are used in advertising, in corporate training — and in schools.
Riparbelli says that humans learn better by watching videos and listening to audio, rather than by reading text. He pointed to people opting to watch tutorials on YouTube or listen to a podcast rather than reading. And as AI gets more advanced, he says, video content will become easier to produce at scale.
Eventually, Riparbelli says, this means that schools will be able to use these avatars to supplement teachers' lessons — like having a personal tutor for each student. Rather than children reading from a textbook, for example, he imagines every student talking to their own AI avatar about a given lesson. That avatar would then assess how well the student comprehended the material and would share that with the teacher. "It's really like sitting down with a teacher that has infinite time and responds immediately to you," Riparbelli says. "I think that is just absolutely magical."
Will AI avatars make education more equitable?
Riparbelli says this technology will "turbocharge" education because lessons could be catered to each child's interests and tastes. Kids could generate videos of their favorite celebrities or athletes teaching them a lesson. It's all about creating a context that is interesting to each kid, he says.
"My guess would be that the generation that grows up today will be much more informed and much smarter than the generation that grew up 50 years ago," Riparbelli says. "I just find it so difficult to not be extremely optimistic when you see how these technologies will impact the future of learning, how it will be one of the greatest equalizers we've ever seen."
Educators remain unconvinced
Still, not everyone is sold on Riparbelli's vision. "There's no video — AI, robot, not even the videos we use to teach in class — that can replace a teacher," says Dominique Jones, an eighth-grade teacher at Francis C. Hammond Middle School in Alexandria, Virginia. "Students are very dependent on human interaction."
During the COVID-19 pandemic, many teachers, including Jones, relied heavily on video lessons. "Did it work?" asks Jones. "I had never had so many F's in my grade book in all of my career."
Jones worries that with the rise of AI, students will suffer from a lack of human interaction and lose important analytical skills. "They're looking for fast, quick answers to everything," she says. "It will become a very disengaged environment."
But as AI becomes more pervasive, Riparbelli says, the question may not be how to avoid it but how to coexist with it. "Let people play around with it. Let them tell you what they do with it. What does it work for? What does it not work for?" he says. "And we don't learn that from reading books about how AI could go wrong. … We learn by doing."
This segment of TED Radio Hour was produced by Rachel Faulkner White and edited by Sanaz Meshkinpour.
The digital story was written by Rachel Faulkner White.
You can follow us on Facebook (TED Radio Hour) and email us at TEDRadioHour@npr.org.
Copyright 2025 NPR
