Generative AI is coming for the classroom, whether teachers like it or not. Here's why many in education think it should be embraced rather than shunned.

• Fears of generative AI helping students cheat are rampant, and some school districts are banning it.
• But concerns over cheating with AI don't address the underlying issues in education, experts say.

Leah Henrickson is no stranger to artificial intelligence.

The digital-media lecturer at the University of Leeds has been using AI in her classroom since 2018 by having her students "write" essays with text generators that can respond to different personalized prompts. After the model comes up with an answer, students break down these blocks of text and discuss the AI-generated arguments.

By approaching AI as a tool with her students, it compels her class to think about the possibilities and limitations of the technology to complete different creative tasks.

Complimentary Tech Event

Transform talent with learning that works

Capability development is critical for businesses who want to push the envelope of innovation.Discover how business leaders are strategizing around building talent capabilities and empowering employee transformation.Know More

"We're getting our students to critically think about these tools," she said.

Before ChatGPT burst onto the scene late last year — making generative AI a household term — some teachers had already been utilizing AI technology in the classroom to get students excited about learning, rather than having them discover it as a tool to cheat on writing assignments.

Edtech startups like Koalluh, which generates short stories to get young kids excited about reading, or Pressto, a tool to help teachers with AI-generated lesson prompts to motivate students to write more, have been around for longer than OpenAI's ChatGPT.

But what sets ChatGPT apart from other AI technologies is the massive potential for students to use it to cheat and plagiarize. It's even spooked the New York City Department of Education — the largest public-school system in the US — which earlier this month announced it was banning ChatGPT on school computers and networks. More school districts across the country are enacting similar policies.

Whether or not such bans are practical or would be effective remains to be seen. Recently, OpenAI CEO Sam Altman said in an interview that the company will develop ways for schools to detect AI plagiarism. OpenAI did not respond to a request for comment.

Insider spoke with multiple AI researchers and academics, startup founders, and education-nonprofit leaders to learn how they're approaching generative AI in the classroom. All of them said that bans on the new technology were misguided at best and that students could circumvent them if they wanted to.

Instead, the fears around generative AI in the classroom point to larger problems with the way local and state governments have been assessing learning in schools in the first place, which tend to focus on rote memorization rather than deeper understanding.

In countries like the US and United Kingdom, education has become just a way to pass a test or write a formulaic essay for a passing grade, and that has "detracted from the core function of education, which is to develop understanding and the process of learning about the world," Henrickson said.

"As long as we're focusing on education as outputs, the GPTs and AI will be a threat, but it can't replace the process of problem-solving, which is the real way to learn," she added.

Welcoming AI in the classroom with open arms

In the long term, embracing the new technology rather than banning it would better prepare young learners for the working world they'll be entering after they leave school, since AI will only become more prevalent in our daily lives, some academics, founders, and education-nonprofit leaders said. If used effectively, generative AI tools like ChatGPT can help students develop critical-thinking and problem-solving skills, which are crucial learning outcomes for most teachers.

Teaching students at all stages how to work with AI will be crucial for their success after finishing school, said Jon Choi, a law professor at the University of Minnesota. He recently ran an experiment to determine whether, if prompted, ChatGPT could provide answers to successfully pass a law-school exam while being blind graded alongside exams taken by real students. And it did.

"My take is that lawyers will begin using these tools if they haven't already, and that law-school courses should be modified to take that into account," Choi said.

Edtech startups like Pressto have developed generative-AI technology as an aid for teachers. The company, which was founded in 2021 by the journalist Daniel Stedman, helps teachers create writing-assignment prompts from traditional essays to news articles or do-it-yourself magazines for students from third through eighth grade.

For students, it may be a fun experience to make a magazine, print it, fold it up, and show it to their family, Stedman said. Since the creative prompts engage the students and are fun, it can increase their learning outcomes, he added.

"We give AI to the teachers and don't encourage it for the students," he said.

Cheating fears

Despite the buzzy reports of generative AI used to score passing grades on assignments, experts agree that most teachers would be able to tell if a student was using the technology to write their papers without at least some effort to cover their tracks.

Richard Culatta, a former high-school Spanish teacher who now runs the International Society for Technology in Education, said that well-trained teachers should be able to spot assignments written only with ChatGPT or another generative AI tool, because they'll recognize an individual student's writing style and see how it's evolved over time.

"If there's a paper being handed in and I can't tell if it was written by a kid or AI, there's a process that's been broken," he said.

And in Choi's law-school-exam experiment, ChatGPT scored a C-plus average and "bombed" any question involving math, he tweeted.

"It would potentially pass law school, but it would be one of the weakest students," he told Insider.

In the short term, a good way to get around cheating fears would be to require students to work on more in-person assignments, said Hadi Partovi, the founder of the nonprofit computer-science-education organization Code.org. One way to teach complex topics could be through in-class debates rather than essays. That way, students are actively engaged and wouldn't be able to use generative AI to complete a full assignment, he said.

But moving the education system forward toward this model is an uphill battle that has only been exacerbated by pandemic learning loss and teacher shortages.

"These are good moments for self-reflection, to make sure education is really doing what we want to do," said Henrickson, the professor at the University of Leeds. "If we reemphasize education as a process rather than a series of outputs, we can use these models to enhance learning."