Cambridge: A groundbreaking Harvard University study has revealed that students learn more effectively and, in less time, when taught through a specially designed artificial intelligence (AI) tutor compared with conventional in‑class active learning.
The randomised controlled trial, published in Scientific Reports, involved 194 undergraduates enrolled in Harvard’s introductory physics course. Over two weeks, each student experienced both teaching methods, enabling researchers to directly compare outcomes.
Better Scores in Less Time
Students using the AI tutor achieved a median post‑test score of 4.5, compared with 3.5 among those taught through active learning sessions. Median learning gains in the AI group were more than double those recorded in the classroom group, with the difference statistically significant.
The AI tutor also proved more time‑efficient. While classroom learners spent about 60 minutes on lessons after accounting for testing, the AI group’s median study time was 49 minutes. Around 70 per cent of AI‑taught students completed the material in under an hour.
Engagement & Motivation Boost
Students reported higher levels of engagement and motivation with the AI tutor. On a five‑point scale, the AI group recorded an average engagement score of 4.1 compared with 3.6 for classroom learners. Motivation scores were also higher, averaging 3.4 versus 3.1.
Implications for Education
Researchers suggested that AI tutors could be deployed to introduce complex material before class, allowing classroom sessions to focus on higher‑order skills such as advanced problem‑solving, projects and collaborative work. They also highlighted potential applications in homework support, exam preparation and remedial learning.
Balanced Perspective
The study’s authors cautioned that AI tutoring may not outperform classroom learning in every context, particularly when lessons demand synthesis of multiple concepts or advanced critical thinking. Nonetheless, the findings underscore AI’s growing potential to reshape education by enhancing efficiency, engagement and outcomes.