[2604.03246] Personalized AI Practice Replicates Learning Rate Regularity at Scale

Computer Science > Computers and Society arXiv:2604.03246 (cs) [Submitted on 9 Mar 2026] Title:Personalized AI Practice Replicates Learning Rate Regularity at Scale Authors:Jocelyn Beauchesne, Christine Maroti, Jeshua Bratman, Jerome Pesenti, Laurence Holt, Alex Tambellini, Allison McGrath, Matthew Guo, Sarah Peterson View a PDF of the paper titled Personalized AI Practice Replicates Learning Rate Regularity at Scale, by Jocelyn Beauchesne and 8 other authors View PDF HTML (experimental) Abstract:Recent research demonstrated that students exhibit consistent learning rates across diverse educational contexts. We test these findings using a dataset of 1.8 million (366k post-filtering) student interactions from the digital platform Campus AI providing further evidence to the observation of regularity in learning rate among students. Unlike prior work requiring manual cognitive modeling, Campus AI automatically generates Knowledge Components (KCs) and corresponding exercises, both of which are validated by human experts. This one-to-many mapping facilitates the application of Additive Factors Models to measure learning parameters without complex cognitive modeling. Using mixed-effects logistic regression, we confirmed the core finding of prior work: students displayed substantial variation in initial knowledge ($\text{IQR} = [2.78, 12.18]$ practice opportunities to reach 80% mastery) but remarkably consistent learning rates ($\text{IQR} = [7.01, 8.25]$ opportunities). Furtherm...