Support Or Substitution: How Chatgpt Affects Cognitive Load, Ownership, and Performance In Problem-Solving
Abstract
Generative artificial intelligence is reshaping knowledge work by enhancing performance while raising
concerns about user agency and ownership. This research-in-progress examines how ChatGPT access
and additional prompt guidance affect task performance, temporal and mental demand, and
psychological ownership in problem-solving tasks. Drawing on cognitive load theory and psychological
ownership theory, we argue that ChatGPT can support performance by offloading cognitively
demanding subtasks, while potentially reducing users’ sense of ownership over the outcome. In a
between-subjects online experiment (n = 136), participants were assigned to one of three conditions:
no ChatGPT, ChatGPT-only, or ChatGPT with additional prompt guidance. Results show that both
ChatGPT conditions improved task performance compared to the control group. ChatGPT access
reduced temporal demand, but neither condition significantly affected mental demand, and no indirect
effects were found. Both experimental conditions reduced psychological ownership. AI literacy
moderated the effects on temporal demand and, partially, on psychological ownership.
Keywords
Generative Artificial Intelligence; Human-AI Collaboration; Cognitive Load; Psychological Ownership