Artificial intelligence learning environments and educational psychology (2016-2025): a systematic bibliometric synthesis and review.

The rapid integration of Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs) into educational environments has fundamentally transformed instructional design building upon earlier AI research in educational psychology. However, there is growing concern that computational innovation is outstripping pedagogical and psychological coherence. To evaluate the intersection of AI learning environments and educational psychology, this study provides a systematic bibliometric synthesis of peer-reviewed scholarship published between 2016 and 2025. Following PRISMA 2020 guidelines, we employed a dualpronged approach, applying multiple descriptive and analytical modeling techniques to analyze a synthetic corpus of 34 systematic review papers and an empirical corpus of 1,376 original research articles. Longitudinal results reveal an unprecedented "generative turn" and a shift in methodological maturity, moving from exploratory computational modeling to rigorous experimental and quantitative validation. Despite these advancements, our analysis identifies a persistent "theoretical gap." While the synthetic literature advocates for deep psychological grounding, empirical studies remain overwhelmingly anchored in surface-level motivational outcomes, failing to adequately operationalize complex constructs such as metacognition, epistemic beliefs, and self-regulated learning (SRL), just to name a few. We conclude that for AI to foster effective, equitable instruction and learning, it must transcend being a mere content delivery mechanism and instead function as a "psychological catalyst." By prioritizing human-centric architectures, cognitive scaffolding, identity, and critical AI literacy, this synthesis provides a structural roadmap for stakeholders to align rapid technological strides with the foundational science of learning.