Can AI make students better problem solvers, and reduce cognitive effort? A recent study by Urban et al. (2024) puts ChatGPT to the test in a complex creative problem-solving experiment.
While AI-assisted students created more creative solutions using less effort, they struggled to accurately assess their own performance. As AI tools reshape education, the key challenge lies not just in using them—but in learning how to think critically alongside them.
Plagiarism concerns
Since ChatGPT was launched for public use in November of 2022, one of the main critiques of the platform has been its impact on originality. This has been especially prevalent in academia, where concerns regarding plagiarism and AI written work have proliferated (Lin et al., 2023). As usage of ChatGPT has progressed, research has emerged surrounding these concerns.
Specifically, Urban et al., (2024) have conducted exploratory research into ChatGPT’s impact on creative problem-solving in undergraduate university students. Published in Computers & Education, the article explored the effect of ChatGPT on students’ ability to solve tasks that are inherently open-ended and rely on the use of creativity and critical thinking skills, such as essay writing or case studies.
Prior to experimentation, a baseline originality measure was implemented (Urban et al., 2024). Participants were randomly assigned to the experimental group that was allowed to use ChatGPT, or to the control group that had to solve the task without Chat GTP. Participants in both groups completed the same complex problem-solving task.
Toying with participants
The task involved a hypothetical scenario in which participants had to imagine they were consulting a toy manufacturer in improving product design. Participants’ responses were analyzed by independent experts for their originality and quality, as well as for self-efficacy.
The results showed that the usage of ChatGTP increased scores for self-efficacy, quality, and elaboration. Crucially, participants allowed to use ChatGTP reported that they found the task less cognitively demanding and effortful, although they did not find that ChatGPT made the task more engaging.
Interestingly, students struggled to accurately self-evaluate their performance, with the perceived usefulness of ChatGPT overshadowing their judgement
Interestingly, students struggled to accurately self-evaluate their performance, with the perceived usefulness of ChatGPT overshadowing their judgement. To conclude, the study results are in line with theories of human-AI hybrid intelligence, suggesting that tasks can be most efficiently solved through a combination of human metacognition with Artificial Intelligence usage (Dellermann et al., 2019).
Urban et al.’s (2024) research is especially relevant to neurodiverse learners as reductions in task effort could help combat difficulties neurodivergent learners face surrounding cognitive load and emotional stress. Cognitive load refers to the capacity of our working memory, our ability to hold and utilise information needed for different tasks (Baddeley, 1992).
Growing research suggests that differences in cognitive processes within neurodivergent students lead to deficits in working memory, impeding academic performance
Growing research suggests that differences in cognitive processes within neurodivergent students lead to deficits in working memory, impeding academic performance (Le Cunff et al., 2023). Urban et al.’s (2024) findings that ChatGPT lowers task effort is a promising area to explore in accommodating neurodivergent students.
The full article, “ChatGPT improves creative problem-solving performance in university students: An experimental study” can be accessed here.
References
Baddeley, A. (1992). Working memory. Science, 255(5044), 556–559. https://doi.org/10.1126/science.1736359
Dellermann, D., Ebel, P., Söllner, M., & Leimeister, J. M. (2019). Hybrid Intelligence. Business & Information Systems Engineering, 61(5), 637–643. https://doi.org/10.1007/s12599-019-00595-2
Le Cunff, A-L., Dommett, E., & Giampietro, V. (2023). Neurophysiological measures and correlates of cognitive load in attention‐deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD) and dyslexia: A scoping review and research recommendations. European Journal of Neuroscience, 59(2), 256–282. https://doi.org/10.1111/ejn.16201
Lin, S.-M., Chung, H.-H., Chung, F.-L., & Yu, L. (2023). Concerns About Using ChatGPT in Education. Lecture Notes in Computer Science, 14099, 37–49. https://doi.org/10.1007/978-3-031-40113-8_4
Urban, M., Filip Děchtěrenko, Jiří Lukavský, Hrabalová, V., Filip Svacha, Brom, C., & Urban, K. (2024). ChatGPT improves creative problem-solving performance in university students: An experimental study. Computers & Education, 215, 105031–105031. https://doi.org/10.1016/j.compedu.2024.105031