2026 Theses Doctoral

The Impact of AI Chat Tools on Learning Communities: A Cognitive Science Perspective

Coleman, Joshua K.

This dissertation investigates how the emergence of large language models (LLMs), particularly ChatGPT, is transforming collaborative learning and public knowledge production in online communities. Focusing on five technical subforums within the Stack Exchange network, the study employs a quasi-experimental, pre-test/post-test design grounded in distributed cognition theory to examine linguistic and behavioral changes before and after the public release of ChatGPT (pp. 28-47).

Quantitative findings indicate a statistically significant increase in question complexity post-release, as measured by higher Flesch-Kincaid Grade Levels across the dataset (pp. 55-59). No significant differences were observed in Lexical Density (proportion of content words), suggesting that syntactic complexity increased without a corresponding rise in semantic information load (pp. 56-59). Across the two nine-month sampling windows, overall question volume declined by 15.6% (75,923 pre vs. 64,072 post), with decreases concentrated in Computer Science, Data Science, and Mathematics, while Academia rose modestly and Artificial Intelligence remained near-flat.

In the answered-question analytic sample (N = 97,513), user engagement metrics remained relatively stable overall, though some communities experienced localized declines in question resolution (accepted-answer rate) and response activity (number of answers per question) (pp. 48-56, Table 8). These patterns support the hypothesis that users offload simpler queries to ChatGPT and use public forums for more complex or ambiguous problems (pp. 59-64). The final chapter contextualizes these findings within broader concerns about educational equity, platform governance, and the future of digital commons (pp. 72–76).

Keywords: artificial intelligence (AI), distributed cognition, large language models (LLM), question complexity, epistemic infrastructure, collaborative learning