Effects of generative artificial intelligence on cognitive effort and task performance: study protocol for a randomized controlled experiment among college students.
Level 5 - mechanism / opinion, no new human data
Study protocol for a randomized controlled trial without reported results.
PubMed 40646586 · doi:10.1186/s13063-025-08950-3
What was done
This paper describes the protocol for a randomized controlled laboratory trial designed to recruit 160 college students (aged 18–35 years) at a German university. Participants will be randomized to complete an analytical writing assignment either with the assistance of a generative AI tool (intervention group) or without it (control group). Cognitive effort will be monitored during the task using eye-tracking (pupil dilation, eye movements) and functional near-infrared spectroscopy (fNIRS) to track brain hemodynamic responses. Subjective perceptions and attitudes toward generative AI will be evaluated via survey.
What was found
No findings or numerical results are reported; this publication is a study protocol.
Why it matters
As generative AI becomes widely adopted in education and professional settings, objective neurophysiological data using fNIRS and eye-tracking may help clarify whether AI tools augment cognitive capabilities or lead to cognitive offloading during complex writing tasks.
Limits
No experimental results are available yet. The planned design evaluates an acute, single-session hypothetical writing task in a laboratory setting among young adult university students, which may not capture long-term cognitive adaptations or diverse non-academic populations.