Unconscious Plagiarism and Artificial Intelligence

JADE DURKEE and MEGAN H. PAPESH, University of Massachusetts Lowell

Fri. Nov 20, 6:00 - 7:30 Poster Session (Poster V-135)

Cryptomnesia, or unconscious plagiarism, is a source memory error where individuals believe they created a new idea that was encountered previously. Although prior research suggests people are more likely to unconsciously plagiarize from non-human sources (e.g., computers), research exploring whether collaborative learning with artificial agents produces similar source memory differentiation errors is sparse. The present study aimed to fill this gap, expanding cryptomnesia research in the age of AI and furthering our understanding of how AI impacts cognitive processes related to source monitoring and memory. We examined how AI as a memory source influenced rates of unconscious plagiarism in a puzzle-solving paradigm with Boggle puzzles. Participants were randomly assigned to solve puzzles with a human or AI partner. Once assigned, participants completed a series of Boggle puzzles with their assigned partner. Each puzzle had three stages: generating solutions, recalling solutions, and finding new solutions. We measured how often participants recalled a partner’s solution as their own or repeated their partner’s solutions, believing them to be “new.” Results will be discussed in terms of both theoretical and applied consequences.