Students Incorporated logo Students Incorporated

Education

Most Schools Skipped the AI Rules. This One Didn't

Nationally, most schools let students use AI with no written policy. In 2024, ICS Bangkok's headmaster said the rules had to come before the access.

Most Schools Skipped the AI Rules. This One Didn't

When the Students Incorporated podcast opened its third season in August 2024, the team behind it made a call that looks better with every passing year: they named artificial intelligence the single biggest trend reshaping education, ahead of nano-learning, hybrid classrooms, and social-emotional learning. Two years on, the data backs them up, and so does their school's position at the time. While most American schools were letting generative AI spread through classrooms with no formal policy at all, the International Community School of Bangkok's headmaster, known on the show as Mr. Luke, described an institution trying to write the rules before it handed out the access.

That gap between adoption and policy hasn't closed. It's gotten more visible. A Fortune report published this month found that 84% of students now use AI for homework, while only about three in ten schools have any rules governing it. A UNESCO survey found fewer than 10% of schools worldwide had adopted formal generative AI guidance, and of those that had, 40% said the guidance was communicated verbally rather than in writing. Set against that backdrop, what ICS was describing in 2024, generative AI opened for specific teacher-directed uses, paired with written policy before wider rollout, was closer to the exception than the norm.

"We Don't Want to Get the Wagon Before the Horse"

Mr. Luke didn't describe the school's approach as cautious so much as deliberate. Asked how ICS was using educational technology, he said the school had opened ChatGPT to students the previous year, but only for teacher-directed activities, alongside policies built specifically for generative AI tools. "We're moving, I don't want to say slowly, but I would say that we're moving very intentionally and methodically to adopt AI as a tool that can enhance education," he told co-hosts Mia and Frank. "We don't want to get the wagon before the horse."

That sequencing (policy first, wide access second) is the piece of the episode that reads differently now than it would have on release day. Adoption moved fast enough since that, for most schools, policy became something to catch up to rather than build ahead of time. ICS, on this account, tried to do it in the other order.

The Trend List Held Up

Before the interview, co-hosts Mia and Frank ran through a top-10 list of 2024 education trends, from number 10 to number one: global competency, sustainability education, personalized learning paths, 21st-century skills, data-driven instruction, social-emotional learning, classroom tech like VR and gamification, micro- and nano-learning, hybrid models, and AI integration at the top. It's the kind of list a class project produces every year, and most such lists don't hold up. This one mostly has, at least at the top: AI adoption in schools has grown fast enough since 2024 that the College Board and RAND data below treat it as close to the default rather than the exception.

Mr. Luke leaned into the adaptive piece specifically when asked where he expected the most change. He compared it to the video games of his own childhood, ones where a player had to demonstrate a skill to unlock the next level, describing modern adaptive-learning software as "obviously a super advanced version of that," adjusting content difficulty in real time as a student demonstrates competency. He also pointed to individualized AI tutoring as a way to fill gaps schools can't staff for directly, noting some universities already use similar tools to backfill prerequisite courses for incoming students missing coursework.

He was describing, without using the term, what the industry now generally calls adaptive learning: software that reads a student's answers in real time and reshapes what comes next, rather than moving an entire class through the same material on the same schedule. It's the same idea researchers have pointed to as one of AI's more durable contributions to education, less flashy than a chatbot that writes essays, but closer to what actually changes how a struggling or advanced student experiences a lesson.

The Soft-Skills Bet

The episode's sharpest moment came when co-host Frank read a quote he credited on the show to a "technology expert" identified as Dr. Trace: that people won't lose their jobs to AI so much as to other people who know how to use it. That attribution is the show's own and couldn't be independently confirmed; the line is one of the most repeated in AI-era career advice, echoing a formulation associated with economist Richard Baldwin's remarks at the World Economic Forum, though a viral 2024 social post also claims to have originated it, and its precise source remains disputed online.

Whoever said it first, Mr. Luke's answer to it has aged into something close to consensus. He argued that AI access itself won't be the dividing line between students who thrive and students who don't, because access is becoming universal. What won't be universal, he said, is the ability to think critically, solve problems, and work with other people.

"Everyone has access to AI, but not everyone has those soft skills."

That claim now has real numbers behind it. The World Economic Forum's Future of Jobs Report 2025, drawn from a survey of employers representing more than 14 million workers across 55 economies, found that resilience, flexibility, leadership, and social influence rank among the skills employers expect to matter most through 2030, alongside creative thinking and curiosity. The same report projects that 39% of current core skills will transform or become outdated by 2030, and that 85% of employers plan to prioritize workforce upskilling in that window. The message from the world's largest employer survey is close to word-for-word what Mr. Luke told two high schoolers in a podcast studio: the technology will be everywhere, and the differentiator will be what humans do with it.

What the Classroom Data Shows Since

The picture that's emerged since 2024 complicates the "slow and steady" framing a little. Research from the College Board found that the share of high school students using generative AI for schoolwork rose from 79% to 84% in just the first five months of 2025, with 69% specifically naming ChatGPT. A College Board follow-up report, based on more than 10,000 survey responses from students, parents, and teachers collected between mid-2024 and mid-2025, found private schools are more than twice as likely as public schools to both permit generative AI use and have written policy governing it. That puts an independent, tuition-funded school like ICS on the more-resourced side of a divide that shapes how differently students encounter the same technology depending on where they're enrolled. RAND's American School District Panel has separately tracked a rise in the number of U.S. districts formally training teachers on AI tools, evidence that policy infrastructure is catching up, unevenly, across the public system.

None of that guarantees any given policy gets the details right. But it does mean the instinct Mr. Luke described on air two years ago, treat the rules as part of the rollout rather than an afterthought, tracks with where the research has landed since. Nano-learning, the trend the co-hosts ranked third, drew the most skeptical response from him on the show: he called it useful but incomplete, noting that some students still learn best from a teacher talking through a subject for forty minutes, and that "there's no silver bullet when it comes to the perfect way to learn." That's held up too. The College Board and RAND data both point toward AI as one tool layered into classrooms already juggling several teaching models, not a replacement for any of them.

The episode closed with device-usage advice that had nothing to do with school policy at all: no phones at the dinner table, no screens in bed, a headmaster admitting he breaks his own rule more than he'd like. It's a small, human coda to a conversation mostly about institutions and infrastructure, and it lands as the same argument in miniature. The tools keep getting more capable. What people do around them, at the dinner table or in a classroom, is still the part worth getting right on purpose.

← Back to all posts

Our sponsors

New episodes every week. Don't miss one.

Subscribe now