Teenagers still want careers in technology, but they no longer believe there’s only one route to get there. With AI poised to automate many coding tasks—and no AP course in “vibe coding”—students are rethinking what skills truly matter. Their teachers are scrambling to keep pace.
“There’s been a shift from taking as much computer science as possible to trying to fit in as many statistics courses as they can,” says Benjamin Rubenstein, assistant principal at Manhattan Village Academy in New York. After two decades in city classrooms, Rubenstein has watched the once-linear “STEM pipeline” splinter into a network of branching paths. For many of his students, statistics feels more practical.
Forty years ago, kids inspired by NASA dreamed of becoming physicists or engineers. Twenty years later, the lure of Google and other tech giants drove them into computer science. Now, AI is reshaping those ambitions, pushing students away from skills machines excel at (coding) and toward areas where human judgment still matters. As interest in computer science degrees dips, STEM-oriented high schoolers are gravitating toward fields that combine computing with analysis, interpretation, and data.
Rubenstein still requires every student to take at least one computer science course “so they understand what’s happening behind the scenes.” But the school’s math department now connects data literacy to real-world purpose: an Applied Mathematics class where students analyse NYPD data to propose policy changes, and an Ethnomathematics course exploring the links between math, culture, and identity. “We don’t want math to feel disconnected from real life,” he says.
It’s a small but telling shift—and not an isolated one. After years of rapid growth, the computer-science boom in higher education is cooling. According to the Computing Research Association, degrees in computer science, computer engineering, and information fields in the US and Canada fell about 5.5 percent in the 2023–2024 academic year.
The appetite for data is clear at the high school level as well. In 2024, 264,262 students registered for AP Statistics—making it one of the most popular AP exams, according to Education Week. AP computer-science exams still draw large numbers—175,261 students took AP Computer Science Principles, and 98,136 took AP Computer Science A—but the trend is unmistakable: data literacy now stands beside coding as an essential skill.
“Students who see themselves as STEM-oriented will pursue whatever they believe makes them valuable in the job market,” Rubenstein says. “The workplace can shift K–12 education simply by signalling what it needs.”
Teachers, meanwhile, are navigating the uneasy reality that AI is both the future they must prepare students for and a tool that, if misused, can short-circuit learning.
Still, Rubenstein believes AI could become a powerful partner rather than a threat. He envisions classrooms where algorithms flag which students understand a concept and which need more support, or suggest tailored data projects based on a student’s interests—tools that make learning more personalised and applied.
It reflects the broader transformation he sees in his students: a shift toward learning how to interpret and use technology, not just create it. Other educators are thinking similarly, exploring how AI tools might deepen data literacy and expand access to individualised STEM instruction.
At the University of Georgia, science education researcher Xiaoming Zhai is already experimenting with what that future might look like. His team is developing “multi-agent classroom systems”—AI assistants that interact with teachers and students to model scientific inquiry.
Zhai’s work targets a new type of literacy: not just knowing how to use AI, but knowing how to think with it. He tells the story of a visiting scholar who had never written a line of code yet created a fully functional science simulation using generative AI.
“The bar for coding has been lowered,” he says. “The real skill now is integrating AI into your discipline.”
Zhai argues that AI isn’t a collection of STEM subjects—it’s becoming central to all of them. He predicts future scientists will use algorithms the way earlier generations used microscopes: to spot patterns, test hypotheses, and push the limits of knowledge. Coding is no longer the frontier; the new frontier is learning how to interpret and collaborate with machine intelligence.
As chair of a national committee on AI in science education, Zhai is working to make that shift explicit. He wants schools to teach students to leverage AI’s precision while staying mindful of its blind spots.
“AI can do things humans can’t,” he says, “but it can also fail spectacularly outside its training data. We don’t want students who think AI can do everything—or who fear it entirely. We want them to use it responsibly.”
That balance—between fluency and scepticism, ambition and identity—is quietly reshaping what STEM means in classrooms like Rubenstein’s. Computer-science classes aren’t disappearing, but they now share the stage with forensics electives, sci-fi-inspired design labs, and data-ethics debates.
“Students can’t think in silos anymore,” Rubenstein says. “You need multiple disciplines to make good decisions.”
AI isn’t coming—it has arrived. Today’s STEM students aren’t resisting it; they’re learning to analyse it, question it, and harness it. The essential skill is no longer writing the code, but understanding the logic well enough to steer the machine.
