Abstract: The rapid integration of artificial intelligence (AI) into online higher education has produced conflicting evidence regarding its impact on higher-order learning. Guided by the Community of Inquiry (CoI) framework and Human-Centered AI–Teaching Presence (HC-AI-TP) theory, this study examined whether the effects of AI-supported instruction on cognitive presence and perceived learning are conditional on teaching presence. Using a cross-sectional quantitative design, survey data were collected from 200 students enrolled in fully online courses. Previously validated instruments were employed, including the Human-Centered AI–Teaching Presence (HC-AI-TP) scale for AI-supported instruction, the CoI Teaching Presence and Cognitive Presence scales, and a validated perceived learning measure. Internal consistency reliability for all instruments was re-established in the present sample prior to hypothesis testing. Pearson correlation analyses indicated strong positive associations among teaching presence, cognitive presence, and perceived learning (r = .55–.68, p < .001), while AI-supported instruction demonstrated a modest bivariate association with cognitive presence (r = .29, p < .001). Regression analyses showed that AI-supported instruction did not independently predict cognitive presence (β = .07, p = .170), whereas teaching presence was a strong positive predictor (β = .49, p < .001). Moderation analysis revealed a significant AI × Teaching Presence interaction (β = .21, p < .001), indicating that AI enhanced cognitive presence only under conditions of strong teaching presence. In contexts of low teaching presence, AI use was associated with diminished cognitive engagement. Mediation analysis further demonstrated that cognitive presence fully mediated the relationship between ethical AI use—operationalized as AI-supported instruction governed by high teaching presence—and perceived learning (indirect effect = .34, 95% CI [.25, .45]). Phase-level analyses showed that the moderating effect of teaching presence was strongest for the integration (β = .27, p < .001) and resolution (β = .31, p < .001) phases of cognitive presence. Demographic variables were non-significant, and robustness checks confirmed the stability of the findings. Overall, the results indicate that AI is neither inherently beneficial nor harmful; its educational value depends on instructional governance through teaching presence. The study advances Bull’s (2026) (HC-AI-TP) theory by positioning cognitive presence as an ethical learning outcome and provides empirical guidance for responsible AI integration in online higher education. Keywords: artificial intelligence in education; teaching presence; cognitive presence; Community of Inquiry; ethical AI; online learning. Title: Ethical Use of Artificial Intelligence and Cognitive Presence in Online Learning: A Human-Centered AI–Teaching Presence (HC-AI-TP) Perspective Author: Dr. David Augustine Bull International Journal of Computer Science and Information Technology Research ISSN 2348-1196 (print), ISSN 2348-120X (online) Vol. 14, Issue 1, January 2026 - March 2026 Page No: 45-65 Research Publish Journals Website: www.researchpublish.com Published Date: 24-January-2026 DOI: https://doi.org/10.5281/zenodo.18360185 Paper Download Link (Source) https://www.researchpublish.com/papers/ethical-use-of-artificial-intelligence-and-cognitive-presence-in-online-learning-a-human-centered-aiteaching-presence-hc-ai-tp-perspective