Artificial intelligence programs might know more than your average regular tutor, and they can definitely help kids improve their learning when they’re being tutored, but a new study finds that they can’t replace the tutor by themselves. It’s another example of where human beings still need to be involved, even as the technology has the ability to help improve academic performance.
That’s from a new briefing by Stanford University’s SCALE — Systems Change Advancing Learning and Equity —initiative, which looked at how AI is being used to help with tutoring.
“In a landscape of reduced school funding, district leaders are looking for ways to save money, and it can be tempting to swap human tutors for AI,” admitted Chris Agnew, the managing director of the AI Hub for Education, which is part of Stanford’s SCALE initiative.
He said years of research show that a human being working as a tutor offers engagement and motivation, and that it ultimately leads to better academic outcomes. Agnew describes that as “high-impact tutoring.” When you already have that, research shows that AI can help drive more gains.
“Imagine a human tutor engaging with a student, and then the artificial intelligence supporting the tutor with additional data about the student, with maybe suggestions on things to say next, grounded in good pedagogy,” said Agnew. “There is emerging evidence — but results are still early — that AI tutoring with human support shows early promise, but we need to learn a lot more.”
Agnew said AI can help identify where gaps in student learning and understanding are, and offer up some instructions on how the human can fill those gaps.
But when you just put a student in front of a computer with an AI tutor, and there’s no human to help supervise, the outcomes aren’t as strong. You’ve probably heard the saying “you can lead a horse to water but you can’t force it to drink.” The same mantra applies here.
“Almost half of students, when provided with an AI tutor, just didn’t use it,” said Agnew, referring to an earlier study SCALE conducted this year. “So they don’t have that emotional connection to provide the persistence to push through when learning gets hard.
“They bring the ability to connect with kids, to engage young people,” explained Agnew. “Then the AI can support that human tutor and bring deeper expertise in math, or deeper expertise in teaching skill.”
The results of this study come at a time when school systems are being inundated with pitches for various programs that aim to help students improve learning, as budgets are getting squeezed and finding enough people who can help students learn is getting harder. The study argues that AI plays no role in maximizing available resources.
“There are real opportunities to increase impact by supporting tutoring, enabling tutoring with AI,” said Agnew. “But … our evidence shows that will be most effective when it’s human tutors (working directly) to kids or human tutors supported by AI.
“There’s a real opportunity where AI can extend the reach of high-impact tutoring and improve learning outcomes in the realm of expanding access, lowering cost, and personalizing practice,” he added. But when the lesson gets hard and students need that extra push, “technology alone can’t replace this student-tutor relationship that really, based on research, we know drives learning gains.”
Get breaking news and daily headlines delivered to your email inbox by signing up here.
© 2026 WTOP. All Rights Reserved. This website is not intended for users located within the European Economic Area.
