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

EXPLAINERArtificial intelligence is spreading in classrooms, but its impact on learning depends less on the technology than on how students use it.Hundreds of millions of primary and secondary students are starting their schools for the new academic year this month. But the advent of Artificial Intelligence is changing the students’ learning process.Students and teachers are grappling with the question: what is homework for, when an AI chatbot can finish it in seconds?Some countries have moved fast with the adoption of AI. China requires at least eight hours of AI instruction a year, and in the Gulf, the United Arab Emirates, Saudi Arabia and Qatar have all begun integrating AI into their curricula.AI chatbots can make learning faster and easier. But what happens when students put the technology aside? How much of the learning was theirs, and how much stayed with the machine?Faster homework, weaker examsTo investigate this, David Stromberg of Stockholm University, with Victor Lei and Yanhui Wu of the University of Hong Kong, ran one of the largest studies of its kind, tracking about 27,000 students aged 12 to 18 in China for 30 months.Their working paper combined data on homework, monthly closed-book exams and high-stakes entrance exams across nine subjects.The results reveal a striking gap between doing schoolwork and learning from it.After adopting generative AI, students’ homework scores rose by 18 percent, while the time spent on each assignment fell by about 30 percent, from 64 minutes to 45 minutes on average.But when the same students sat monthly exams without AI, their scores fell by 20 percent within six months. Their performance in high-stakes entrance exams also declined, with the full effect emerging over a longer period.(Al Jazeera)The Organisation for Economic Co-operation and Development (OECD) came to a similar conclusion in its Digital Education Outlook 2026, a synthesis of the emerging research published in January. Where generative AI is neither built to teach nor used under any structure, it found, outsourcing tasks to the model improves performance with no real learning gains.