Faster homework, poor exam results: What AI is doing to students’ learning

Artificial intelligence is reshaping the educational landscape, particularly among students who are using AI tools to complete assignments more quickly, but at the cost of deeper learning, according to recent research. A study conducted by researchers from Stockholm University and the University of Hong Kong tracked nearly 27,000 students aged 12 to 18 in China over 30 months. Their findings suggest that while AI-assisted homework leads to higher grades and reduced time spent on tasks, it does not translate into stronger performance during exams taken without such assistance. The study analyzed data on homework completion, monthly closed-book exams, and high-stakes entrance exams across nine subjects. After the introduction of generative AI, students' homework scores increased by 18 percent, and the time spent per assignment dropped by approximately 30 percent, from an average of 64 minutes to 45 minutes. However, when these same students took monthly exams without access to AI, their scores fell by 20 percent within six months. Performance in major entrance exams also showed a decline, though the full impact became evident over a longer timeframe. This pattern aligns with conclusions drawn by the Organisation for Economic Co-operation and Development (OECD) in its Digital Education Outlook 2026, which synthesized emerging research on AI's role in education. The OECD noted that when generative AI is not designed for teaching purposes or used in structured environments, relying on it to complete tasks can improve immediate performance without fostering genuine learning. Research co-authored by Dragan Gasevic refers to this phenomenon as "metacognitive laziness," suggesting that students may become overly reliant on AI rather than engaging in critical thinking and problem-solving independently. Student usage patterns further highlight the complexity of AI’s influence. A survey by Ipsos for the Vodafone Foundation questioned 7,000 students aged 12 to 17 across seven European countries about how they use AI for studying outside of school and without direct teacher guidance. The most frequent use was seeking information, cited by 56 percent of respondents. An additional 45 percent used AI to explain terms and concepts, while 31 percent relied on it to generate complete solutions to tasks. Fewer students engaged AI in ways that supported self-directed learning, only 29 percent used interactive or adaptive content to guide their studies, and just 20 percent utilized AI to develop personalized learning plans and monitor their progress. Beyond academic work, AI is beginning to shape broader aspects of student life. According to the HEPI/Kortext Student Generative AI Survey 2026, which surveyed 1,054 full-time undergraduate students in the United Kingdom, nearly half (49 percent) believed AI had enhanced their overall student experience. Benefits cited included time savings, improved comprehension, and access to instant, tailored support. The survey also explored the impact of AI on feelings of loneliness. While 59 percent of students reported no change in their sense of isolation, the remaining 41 percent were evenly divided: 21 percent felt less lonely, and 20 percent felt more so. In several regions, the integration of AI into education systems is accelerating. In China, students are required to receive at least eight hours of AI-related instruction annually. Meanwhile, the United Arab Emirates, Saudi Arabia, and Qatar have initiated efforts to incorporate AI into their curricula. These developments reflect a global shift toward leveraging artificial intelligence to enhance educational outcomes, although the long-term consequences remain uncertain. As AI continues to permeate classrooms worldwide, educators and policymakers face the challenge of ensuring that students benefit from the technology without compromising essential learning processes. The balance between efficiency and depth of understanding will likely define the future of AI in education.