–
Jill Barshey for The Hechinger Report
It’s easy to get caught up in the hype about artificial intelligence tutors. But the evidence so far suggests caution.
Some studies have found that chatbot tutors can Backfire Because the students Trust them Too much, get a spoonfed solution, and the material fails to absorb. Even when AI tutors are designed not to answer, they consistently produce better results than old-fashioned learning without AI.
Still, the researchers who produced these skeptical studies haven’t given up hope. Some are still experimenting, trying to create better AI tutors. A promising idea has nothing more to do with how an AI tutor explains the concepts and asks students to practice them next. In this story, The Hechinger Report Examines how researchers are experimenting with new approaches to AI tutoring and personalized learning.
A team at the University of Pennsylvania, which included some AI skeptics, recently A new method has been tested AI tutoring in a study of nearly 800 Taiwanese high school students learning Python programming. All students used the same AI tutor, which was designed to avoid feedback.
But there was one key difference. Half of the students were randomly assigned to a fixed sequence of practice problems, progressing from easy to difficult. The other half received a personalized sequence with an AI tutor that continuously adjusts the difficulty of each problem based on how the student is performing and interacting with the chatbot.
The concept is based on what educators call the “zone of proximal development.” When problems are too easy, students get bored. When they are too difficult, students get frustrated. The goal is to keep students in a sweet spot: challenged but not overwhelmed.
The researchers found that students in the personalized group did better on the final exam than students in the fixed problem group. The difference was pegged as the equivalent of six to nine months of additional schooling, an eye-watering claim for an after-school online course that lasted just five months. AI Tutor’s inventor, Angel Chung, a doctoral student at the Wharton School, admits that his statistical unit conversion is “not a perfect guess.” (A draft paper about the experiment was posted online in March 2026 but has not yet been published in a peer-reviewed journal.)
Still, it’s early evidence that small changes — in this case, calibrating the difficulty of practice problems to the student — can make a difference.
Chung says that ChatGPT’s responses can already feel very personal because they’re directly responding to a student’s unique question. But that level of personalization isn’t enough. “They don’t know what students usually don’t know,” Chung said. “The student lacks the ability to ask the right questions to get the best tutoring.”
To address this, Chung’s team combined a large language model with a separate machine-learning algorithm that analyzes how students interact with the online course platform — how they answer practice questions, how often they correct or edit their coding, and the quality of their interactions with the chatbot — and uses that information to decide which problem to show next.
In other words, personalization is not just about tailoring the explanation. It’s about tailoring the learning path itself.
That idea is not new.
Long before generative AI tools like ChatGPT were invented, education researchers were building “intelligent tutoring systems” that tried to do something similar: guess what a student knew and provide the correct next problem. These earlier systems couldn’t generate natural conversations, but they could provide hints and instant feedback. Rigorous research has shown that well-designed versions help students learn significantly more.
Their Achilles’ heel was engagement. Many students simply do not want to use them.
Today’s AI tools can help solve that problem. Students may feel more interested in a chatbot that speaks to them almost humanly.
In the University of Pennsylvania study, students in the personalized group spent more time practicing, about three extra minutes per problem, adding about an hour per module to the Python course, compared to half the time (half an hour or less) for comparison students. The researchers found that these students did better because they were more engaged in their practice tasks.
Students’ prior knowledge of a topic affects how well personalized sequencing works. Students who were new to Python outperformed those who already had experience with Python, who performed exactly the same sequence of practice problems. Less elite high school students were also found to benefit more.
All Taiwanese students in this study volunteered for an optional computer programming course that could strengthen their college applications. Many were highly motivated with highly educated parents and many already had coding experience.
It’s not clear whether chatbots will work with less motivated students who are behind in school and need extra help.
One possible solution: fusing new and old.
Ken Codinger, a professor at Carnegie Mellon University and a pioneer of intelligent tutoring systems, is experimenting with using New AI model to alert remote human teachers Those who can inspire struggling students who are in flux. “We’re having more success,” Koedinger said.
People are not obsolete – yet.
This story was also published change of mind
This is the story is produced by The Hechinger ReportA nonprofit, independent news organization that covers, reviews, and distributes education Stacker.
–
Previously published at hub.stackernewswire
Good Men Project is pIndustry in a multi-platform ecosystem: flagship sites, Medium publications, substacks, social media. We publish experts, authors, therapists, coaches, activists and wellness professionals. AI tools look for stable, context-rich domains with a clear topical focus. GMP’s mission-driven content helps articles perform better in AI-driven discovery.
If you believe in the work we’re doing at The Good Men Project, please join us as a Premium Member today.
All premium members can watch The Good Men Project without any ads.
Need more information? A full list of benefits is here.
Photo credit: iStock
