The Impact of AI on Student Learning: A Growing Concern
It is difficult to articulate this issue delicately: as our reliance on artificial intelligence increases, our intellectual capabilities may be diminishing. While scientific research has yet to establish a direct causal link between artificial intelligence and a decline in human intelligence, various studies are emerging that suggest this trend. The latest research from China delves into the consequences of university students using AI tools automatically, without critically assessing, questioning, or even comprehending the responses generated by chatbots.
Hui Zhao and Huijuan Gu, researchers from Zhoukou Normal University, conducted a study involving 487 students from four universities in Henan Province. The participants completed a questionnaire addressing four key areas: unreflective use of generative artificial intelligence, self-efficacy, motivation, and self-directed learning. The survey was conducted between June and August 2025, with data analysed using structural equation modelling.
Understanding Unreflective AI Use
The term “unreflective use of generative AI” refers to the behaviour of accepting responses, texts, or ideas produced by AI without critically evaluating or verifying the information. This does not pertain to using tools such as ChatGPT for learning; rather, it involves delegating aspects of intellectual processes without maintaining active engagement. Common practices among respondents included generating ideas for projects, obtaining explanations, refining writing, and solving academic problems.
What were the findings of the study? It was observed that a higher frequency of unreflective AI use correlated with a decreased level of self-directed learning. This latter concept refers to an individual’s ability to initiate, manage, and evaluate their own learning independently. The analogy might be asking ChatGPT to complete your homework, copying it into a document, saving it without a thorough review, and returning to leisure activities. In the statistical model, unreflective usage had a negative impact on this capacity. Furthermore, using AI in this manner was also negatively associated with the two psychological factors studied: self-efficacy (the confidence in one’s ability to learn and solve problems) and motivation.
The Interplay Between AI Use and Learning Efficacy
The researchers conducted a mediation analysis to assess the extent of these relationships and found that unreflective use not only diminishes a student’s independence but can also undermine their confidence and enthusiasm for learning. The negative relationship with self-directed learning was found to be partially mediated by motivation and self-efficacy. A sequence emerged where unreflective usage was linked to lower self-efficacy, which in turn correlated with decreased motivation, ultimately leading to reduced self-directed learning.
Interestingly, the study identified statistically significant differences between male and female participants. The impact of unreflective usage on motivation was more pronounced among males, while the negative effects on self-efficacy and self-directed learning were more evident in females. The authors interpret these differences through the lens of social cognitive theory, although the findings indicate associations rather than establishing gender as a causal factor.
Recommendations for Educational Institutions
The authors do not conclusively assert that AI is detrimental to education. The concern arises when machines replace human independent thinking. Consequently, they recommend that universities educate students on how to critically evaluate and interpret AI-generated responses, and that educators design activities where technology serves as a tool for exploration while also demanding analytical and reflective engagement from students.
However, the study does have significant limitations. The sample was drawn exclusively from universities in Henan and presented an uneven gender distribution, with 80% of participants being female. Additionally, the data collected was cross-sectional, allowing for the identification of statistical relationships but not establishing a causal link. The authors themselves highlight the necessity of future longitudinal studies to determine the long-term effects of unreflective AI use on learning autonomy.
Concluding Observations on AI in Education
The study concludes that “the generative AI’s ability to provide immediate and comprehensive solutions deprives students of opportunities to accumulate mastery experiences through exploration and problem-solving, thereby directly undermining their confidence in their ability to successfully tackle learning tasks.” Moreover, this readily accessible problem-solving approach shifts intrinsic learning motivations—such as curiosity and interest—toward an instrumental pursuit of external results (for instance, completing tasks swiftly), leading to a significant decline in motivation to learn.
The findings of this study can be accessed in the journal Scientific Reports.