Abstract
<p>The exponential rise of artificial intelligence-generated content in recent years has raised questions about how humans process, remember, and recall information based on its assumed source. Our study examined whether labeled authorship affects memory recall by testing the hypothesis that memory for human-authored text would be superior to memory for text believed to be generated by a large language model (LLM). A total of 122 undergraduate college students were randomly assigned to read identical excerpts of contemporary American prose labeled as either human-authored or AI-generated, then completed a surprise memory test along with measures of recall confidence and text readability, likability, and engagement. We observed higher memory recall in participants who believed they were reading a passage written by a human author, while participants who thought the story was written by an LLM recalled significantly fewer details. Participants also reported significantly higher engagement with a human-attributed text, though engagement did not correlate with memory performance. These findings suggest that perceived authorship influences memory recall independently of subjective engagement, with implications for understanding human-AI interactions in educational contexts.</p>