What does the best evidence show? AI tutors can dramatically improve learning while saving time.
The strongest study here — a randomized controlled trial published in 2024 — directly compared an AI tutor to an active learning class (a high-quality teaching method) with college students. The AI tutor was built using the same pedagogical best practices as the lectures. Students using the AI tutor learned more than twice as much content in less time, and they reported feeling more engaged and motivated [1]. This is a striking result: it suggests that, at least for some subjects and age groups, AI can outperform even well-designed human-led instruction.
A separate semester-long study with 51 psychology students taking a neuroscience course reinforces this. The AI tutor automatically generated microlearning questions from course materials and built a personalized model of each student's knowledge. Students who actively engaged with the tutor achieved significantly higher grades — an average improvement of up to 15 percentile points compared to a parallel course without the AI tutor [2]. The tutor's predictions of student grasp also strongly correlated with actual exam grades, showing the system was accurately tracking learning [2].
But does it work as well for younger students? The evidence is more mixed.
When you move from college to K-12 classrooms, the picture becomes less clear. Two systematic reviews (one from 2024, the other from 2025) looked at 28 and 20 studies respectively, covering thousands of K-12 students. Both found that intelligent tutoring systems (ITS) generally had positive effects on learning compared to traditional teaching [3][5]. However, when compared to non-intelligent computer tutoring — basically, simpler software that doesn't adapt to the student — the benefits largely disappeared [3][5]. This is a crucial caveat: the AI advantage may be small or nonexistent if the alternative is already a decent digital tool.
The reviews also noted that most K-12 studies were short (many under a few weeks) and focused on STEM subjects, making it hard to know if the effects last or apply to other areas [3][5]. So while AI tutoring can reduce workload by automating feedback and personalization, its learning benefits for younger students are less certain than for college learners.
What about the catch? Students still want human interaction for some things.
Even where AI tutoring works well, it doesn't replace all human roles. A 2023 study had postgraduate engineering students use a chatbot to help create their personal study plans. The main advantages were scalability and 24/7 availability — real workload reducers for staff. But students reported that the chatbot lacked individuality, didn't inspire them, and couldn't fully replace human counselling [4]. This suggests that for tasks requiring empathy, motivation, or nuanced guidance, human teachers remain essential. The AI can handle the routine, but the human handles the relationship.
About These Sources
This answer is built on 5 peer-reviewed studies — published from 2023 to 2025, 4 from 2024 or later, 2 in Q1 journals, collectively cited 123 times — selected as the most relevant from 5 studies that passed quality screening, drawn from 47 papers retrieved from a database of over 500 million.
Sources used in this answer
AI Tutoring Outperforms Active Learning
In a randomized controlled trial with college students, an AI tutor led to more than double the learning in less time compared to an active learning class, with higher engagement and motivation [1].
Effective learning with a personal AI tutor: A case study
In a semester-long study with 51 psychology students, those who actively used a personalized AI tutor scored up to 15 percentile points higher on exams than a parallel course without the tutor [2].
A systematic review of AI-driven intelligent tutoring systems (ITS) in K-12 education
A systematic review of 28 K-12 studies (4,597 students) found that intelligent tutoring systems generally improved learning over traditional teaching, but benefits were not significant when compared to non-intelligent computer tutoring [3].
Tutoring Postgraduate Students with an AI-Based Chatbot
A study with postgraduate students using a chatbot for study planning found it improved scalability and availability, but students felt it lacked individuality and could not fully replace human counselling [4].
Navigating the Future of Learning: A Systematic Review of AI-Driven Intelligent Tutoring Systems (ITS) in K-12 Education
A systematic review of 20 K-12 studies (2,853 students) found positive effects of intelligent tutoring systems on learning, but these effects were mitigated when compared to non-intelligent tutoring systems [5].