Abstract
<p>Reading primary scientific literature is essential yet challenging for undergraduate science students, who often report low confidence and difficulty navigating complex texts. Guided by Expectancy-Value Theory, this mixed-methods study examined changes in students' motivational beliefs across a biology literature review assignment in which students used Scite, an AI-powered citation analysis tool, including science identity; self-efficacy; and intrinsic, utility, and attainment values and perceived costs of reading primary scientific literature. Students enrolled in a 200-level Foundations of Research in Biology course were assigned a structured literature review on pharmacological interventions for a disease of their choice, using Scite.ai to locate, evaluate, and synthesize primary research. A racially diverse sample of 27 undergraduate biology students (48% women, 42% first-generation college students) completed a posttest questionnaire, including a subsample (n = 21) with matched pretest questionnaire responses. Paired-sample t tests revealed statistically significant, medium-sized increases in intrinsic, utility, and attainment values and science identity, alongside an unexpected rise in perceived costs (d = .51-.65); results revealed a small nonsignificant increase in self-efficacy. Students rated Scite as highly useful for identifying, summarizing, and synthesizing research. Thematic analysis of open-ended responses revealed the salience of perceived utility, enjoyment, cognitive support, and credibility, alongside concerns about reliability, ease of use, and occasional identity-related conflicts. Findings suggest that AI-assisted engagement with PSL carries complex motivational implications, warranting controlled investigation.</p>