A new study reveals that while ChatGPT can improve immediate task performance, it might encourage students to become overly dependent on AI technology. The research, conducted by Shufang Tan and published as an original article, compared different learning support methods among university students.
The randomized experimental study involved 117 university students who were divided into groups receiving support from different sources: ChatGPT, human experts, writing analytics tools, or no additional help. The researchers analyzed the students’ motivation, self-regulated learning processes, and performance outcomes.
While the ChatGPT group showed better improvement in essay scores, there were no significant differences in knowledge retention or transfer compared to other groups. Notably, the study found that AI support might lead to what researchers term “metacognitive laziness,” where students become less engaged in deep learning processes.
The research revealed no differences in post-task intrinsic motivation among the groups, but significant variations appeared in how students approached their learning process. These findings suggest that while AI tools can enhance immediate performance, they may not contribute to longer-term learning goals.
The study recommends that students should focus on using AI to deepen their understanding rather than merely completing tasks efficiently. Teachers are advised to carefully consider which tasks are suitable for AI assistance and develop strategies to maintain active student engagement.