What Schools Are Forgetting in Their Race to Embrace A.I.
I just read two articles that made me reflect on what we are doing in schools today and the way forward. The articles are from the New York Times and the Baille Gifford.
I urge you to read the articles in full and reflect on their findings.
I have chosen some points that I think are important and will try to explain why. I think this should be discussed among teachers and students.
In New York State, phones are now banned in all public schools, from kindergarten to 12th grade. More than two dozen other states have enacted similar policies. The early data on such bans looks pretty ambiguous to me, less like a silver-bullet cure-all for what ails American teens and more as if we’ve achieved what might be a small uptick in well-being and achievement. But I can’t really object to the restrictions, even as a relative skeptic of the smartphone theory of everything. That’s because attention in class is pretty precious and there’s no way the presence of phones doesn’t degrade it, even if only on the margins. Nobody used to tolerate a Walkman or a gaming console in algebra either, and I can’t say I’m surprised that, in those places where phone bans have been implemented, teachers seem pretty elated.
But in many places where phones are being withdrawn and sealed in Yondr pouches, A.I. has infiltrated already. Not all technology is the same, of course, and A.I. tutors represent a different kind of educational promise, at least in theory. But over the past few decades, one dream techno fix has followed the last in the imagination of education reformers, sometimes with great fanfare: laptops in classrooms, tablets, Khan Academy video tutorials or MOOCs. The hope, each time, was that technology could do some of what we’d always relied on scarce humans to deliver, allowing more customization of learning and freeing up overworked teachers to spend their time in more targeted ways. And overall, the returns have been disappointing at best. Some researchers even suggest that it is precisely the arrival of technology in American classrooms, starting about 25 years ago, that best explains recent declines in American academic performance. “This has been the greatest blunder in the past decade of K-12 education,” Molly Worthen wrote in The Times last month — “the decision to give every child a personal computer and to gamify everything from standardized test preparation to recess.” The New York Times
Schools are beginning to recognize that phones damage attention and classroom focus, yet many of the same systems are rapidly introducing AI tools without the same level of caution. The point is not that all technology is harmful, but that schools have repeatedly embraced digital “solutions” before fully understanding their effects on learning. For teachers, this raises an important question: if attention, human interaction, and sustained thinking are precious, then AI should not be adopted simply because it promises efficiency or personalization. It must be judged by whether it genuinely strengthens learning, or whether it becomes another technological shortcut that weakens the habits students most need to develop.
Your mind is changing already.
MIT studied participants’ brains over four months as they wrote essays under three conditions: unaided, with search engines or with AI assistants. The results were stark. AI users showed the weakest connectivity of all three groups, indicating they were barely engaged. By the third session, they had essentially outsourced the entire writing process. When later asked to write without AI, they remembered little of their own earlier essays and showed weaker brain activity associated with deep memory encoding. Most strikingly, 83 per cent of the AI users couldn’t provide a single correct quote from essays they had written just minutes earlier. The effort that builds durable learning had been bypassed entirely. Baille Gifford
This was a paragraph that stood out for me. I think this should be a warning to us all. We are taking the easy way out, and we are paying for the consequences.
There is a deeper problem beyond skill erosion. People tend to trust AI even when they shouldn’t. Psychologists call this automation bias, the tendency to defer to a computer-generated answer over your own judgement, even when your judgement is right. When AI makes confident but incorrect predictions, human performance collapses. A systematic review of 35 studies found that in experimental conditions where AI provided incorrect predictions, radiologist accuracy fell from roughly 80 per cent to as low as 20 per cent for less experienced practitioners, a dramatic illustration of how overreliance on a confident but wrong system can override professional judgement. AI literacy was not sufficient to protect against this effect. Knowing the tool is fallible did not reliably prevent overreliance.
The pattern is not confined to medicine. In 2023, a New York lawyer submitted a legal brief containing six entirely fabricated case citations generated by ChatGPT. When the judge flagged them, the lawyer didn’t check a legal database. He asked ChatGPT whether the cases were real. It assured him they were. He was fined $5,000. Since then, more than 600 similar cases have been documented in US courts alone. Bailliegifford
Examples like these are many. And even if we know this, we keep on using AI. It is easy, accessible and at your disposal 24/7.
A thousand writers, one story
The creativity domain reveals a particularly important dynamic. A meta-analysis of 28 studies found that AI-human collaboration boosts individual creative output compared to working alone. But it simultaneously produces a large homogenisation effect on the diversity of ideas generated. People using AI come up with better individual ideas but more similar ideas to each other. Imagine a thousand writers each producing better stories than they could alone, but all producing essentially the same story. Bailliegifford
I think this says it all to be honest.
With schools, it’s a lot less obvious how exactly we got here in the first place, with A.I. planted so quickly in classrooms and curriculums that parents and advocates are left to try to weed it out. Or at least institute some common-sense guidelines: more public input, more transparency about what products are being used, more restrictions on data collection and student privacy and above all, perhaps, more rigor regarding whether or to what degree the tools being used are actually helpful to student achievement. This isn’t just resistance from the familiar coalition, in other words. It’s parents shocked at how quickly some very new technology has so hastily remade school for their kids. These are 5- and 6-year-old brains we’re now running our Amira experiment on. Most Americans first heard of L.L.M.s less than four years ago, after all. The New York Times
This is important because it asks who decided that AI should enter classrooms so quickly. It is not enough to say that the technology is new and exciting. When schools use AI with young children, parents and teachers should know what tools are being used, what data is collected, and what evidence there is that the tools actually help students learn. We should be especially careful when the students are only five or six years old. At that age, they are not only learning to read; they are also developing attention, confidence, curiosity, and trust in learning. Schools should not rush into using AI simply because the technology is available. Teachers and parents need to be part of the conversation, and the first question should always be: does this help the child learn better?
The result: Even amid remarkable resistance, the technology appears from many vantages to possess the shimmer and the solidity of a fait accompli. We tell ourselves that we know artificial intelligence will remake education and then race to make it so, without asking too many questions about what kind of future we’re actually building. “The obvious rebuttal is no: We actually can’t see the future of A.I.,” Farrell writes. And it’s not just in schools that we may be forgetting that. The New York Times
