acelearning outcomes when university students complete a writing task with ChatGPT support
Source article: Beware of metacognitive laziness: Effects of generative artificial intelligence on learning motivation, processes, and performance
Abstract With the continuous development of technological and educational innovation, learners nowadays can obtain a variety of supports from agents such as teachers, peers, education technologies, and recently, generative artificial intelligence such as ChatGPT. In particular, there has been a surge of academic interest in human‐AI collaboration and hybrid intelligence in learning. The concept of hybrid intelligence is still at a nascent stage, and how learners can benefit from a symbiotic relationship with var…
Contested: both sides are scored from claims and sources, not community votes.
Newspaper reporting and correspondence; a manual for reporters, correspondents, and students of newspaper writing by Hyde, Grant Milnor. Public domain
In a December 2024 randomized lab experiment, 117 university students completed a writing task with support from ChatGPT, a human expert, analytics tools, or no extra support. Researchers measured intrinsic motivation, self-regulated learning processes, and performance, finding no motivation differences but different process patterns and higher essay score gains for the ChatGPT group without higher knowledge gain or transfer.
The findings matter because they separate short-term performance boosts from deeper learning, suggesting that efficiency gains from generative AI may come with reduced metacognitive engagement. Uncertainty remains about whether the observed dependence and process differences persist across tasks, disciplines, and longer-term learning outside the lab.
- Randomized experiment with 117 university students compared support from ChatGPT, human expert, writing analytics tools, and no extra tool on a writing task.
- No difference in post-task intrinsic motivation across groups, but significant differences in frequency and sequences of self-regulated learning processes.
- ChatGPT group improved essay scores more, yet knowledge gain and transfer were not significantly different from other groups.
- Authors warn AI support may promote dependence and metacognitive laziness, hindering deep self-regulation.
Learners supported by ChatGPT showed higher short-term essay score improvement compared to other support conditions.
Learners using ChatGPT may develop dependence on the tool and exhibit metacognitive laziness that hinders self-regulation and deep engagement.
The rundown
The study randomly assigned 117 university students to four conditions for a writing task: ChatGPT support, chat with a human expert, writing analytics checklist tools, and no extra tool, collecting multi-channel data on motivation, self-regulated learning, and performance.
Results showed no group difference in post-task intrinsic motivation, but distinct patterns in self-regulated learning processes, with the ChatGPT group achieving greater essay score improvement without greater knowledge gain or transfer.
The authors interpret the pattern as evidence that ChatGPT can improve short-term task completion while potentially encouraging learners to offload metacognitive monitoring and evaluation, raising concerns for hybrid intelligence design.
Findings come from a single lab-based writing task with 117 university students, limiting generalizability to other tasks, contexts, and populations.
Sources
- Peer-reviewedBritish Journal of Educational Technology2024-12-10
How should this claim be treated?
The debate