From The Morning in the NY Times
Summary:
This article explores how tech companies, including Microsoft and OpenAI, are aggressively integrating AI chatbots into K–12 and college classrooms—offering tools for grading, tutoring, and lesson prep, often with little research backing their educational value. While marketed as essential for future-readiness, the widespread use of AI in schools raises concerns about student data privacy, commercialization of education, and whether these tools truly enhance learning.
Danielson Domain:
Domain 1: Planning and Preparation – especially 1d: Demonstrating Knowledge of Resources and 1e: Designing Coherent Instruction
A.I. in the classroom
Chatbots have wormed their way into everything: phones, cars, TVs, offices. They’re also in kids’ classrooms.
Microsoft and OpenAI announced yesterday that they would spend millions on a new program that will train teachers to use artificial intelligence. It’s part of a bigger push by tech companies to get their chatbots into schools. They’re selling A.I. subscriptions to administrators and promising them that the bots will help teachers grade assignments, prepare lessons and draft recommendation letters. The companies say A.I. proficiency will prepare kids for the work force.
They also approach students directly with discounted subscription rates around exam periods. It’s an old playbook: Get kids hooked, and you’ve got future customers.
But do chatbots actually help them learn? So far, there’s little evidence. Today, I explain how students have become guinea pigs in a national classroom-learning experiment.
What’s happening?
After years of hesitancy and hand-wringing about A.I., schools are starting to experiment with chatbots — some with enhanced privacy guardrails, some without. In a nationally representative survey, nearly half of districts reported having provided A.I. training for their teachers as of last fall. That’s twice the number from the previous year.
In Kelso, Wash., middle and high schoolers used Google’s Gemini this school year for tasks like research and writing. In Newark, an A.I. tool from Khan Academy helps teachers place elementary-school students into study groups based on their skill levels. It also answers students’ questions as teachers give lessons.
Colleges are buying chatbots, too. The California State University System just signed a $17 million deal with OpenAI to give its 460,000 students access to ChatGPT, despite major state funding cuts. The school wants to equip students with A.I. to debug computer code, make digital art, edit essays and research assignments. Schools like Duke and the University of Maryland are among a growing group that have introduced homegrown chatbots for similar tasks.
Same pitch, new era
Tech companies are using an old marketing strategy: Promise that the latest tech will solve classroom problems. In the early 2000s, they told parents and educators that laptops would revolutionize classroom learning. Districts spent millions.
Two decades later, tech companies are still peddling the same fear of missing out: They suggest students need cutting-edge tools for tomorrow’s economy, and schools that don’t provide them are setting their students up for failure. “‘I don’t want my kids to get left behind.’ That’s the first thing we hear from districts,” Vicki Zubovic, who heads outreach for Khan Academy’s new classroom A.I. service, told me.
The government is on board, too. President Trump signed an executive order in April urging schools to integrate A.I. into classrooms at all grade levels. He said doing so would be necessary “to ensure the United States remains a global leader in this technological revolution.”
Will it help students learn?
While tech companies promise that A.I. can facilitate “personalized learning,” many students and educators are simply using chatbots as a sophisticated search engine. (Some also use it to cheat, including by drafting essays.) The Jetsonian features are familiar; interview-prep bots and virtual tutors have been around for years.
Julia Kaufman, who tracks national education data for the RAND Corporation, told me that it was “really hard to know” whether A.I. would actually improve student learning. Since the tools are so new, there’s virtually no research on their efficacy yet.
Laptop programs offer a sobering precedent. They modestly improved students’ long-term achievement: An analysis of 10 studies found “small” but statistically significant bumps in writing, math and science. But those gains often relied on teacher buy-in and revamped curriculums — and fell short of interventions like reducing class sizes and offering tutors.
This time around, the stakes are arguably higher. A generation of students is learning what it means to coexist with — and depend on — powerful, often opaque technology. In many cases, they’re handing over their data to tech companies. And researchers won’t know for years whether the experiment has worked.
