International assessments indicate that 15-year-old students in developed countries have reached their lowest proficiency levels in math, reading, and science since testing began in the early 2000s. The Programme for International Student Assessment (PISA), administered by the Organization for Economic Cooperation and Development (OECD), found that reading scores saw a particularly steep decline, equivalent to more than a year of education, between 2015 and 2025.
In the United States, reading scores fell by approximately 14 percentage points between 2022 and 2025. These findings, which encompass data from over 760,000 students across 91 countries, are closely watched indicators of global educational performance.
While the report points to the lingering effects of the COVID-19 pandemic and a general decline in attention spans as contributing factors, it also examined the growing use of generative artificial intelligence among students. The PISA survey, conducted for the first time since AI's widespread availability, found that 86 percent of students use AI at least occasionally for schoolwork, with about 10 percent using it daily.
The impact of AI on learning appears to be nuanced. Students who reported never using AI generally outperformed those who did. However, weekly users who employed AI to assist in their learning, rather than for completing specific tasks, performed similarly to non-users. Furthermore, frequent AI users who also received instruction in AI literacy scored slightly higher.
This suggests that the effectiveness of AI in education depends heavily on how it is utilized. While some AI tools offer potential benefits, such as reducing teacher workload and providing personalized tutoring through educational chatbots, concerns about its impact have led to policy interventions. For example, New York City has implemented a one-year moratorium on student-facing generative AI tools in its public elementary and middle schools to allow for evaluation of their educational value.
In upstate New York, a federally funded research center is developing AI tools for children with learning and language disorders, with one mother reporting significant improvements in her autistic son's communication skills through an AI reading app. This highlights the complex policy discussion surrounding AI in education, where potential benefits must be weighed against risks and implementation challenges.