• AI technologies • Digital Learning theory • Metaverse, NFTs, Blockchain • eLearning applications

 Constantine Andoniou is an Associate Professor of Education at Abu Dhabi University, specializing in AI-driven education, digital learning, and emerging technologies. He holds a Ph.D. from The University of Queensland and has extensive international academic leadership experience. His research focuses on AI literacy, human–AI cognition, and the future of education in the age of intelligent systems. Reference: The Illusion of Understanding in Advanced AI Systems:Trust, Responsibility, and Machine Cognition Advanced AI systems increasingly produce outputs that resemble reasoning, explanation, and analysis. In high-stakes domains such as cybersecurity, fraud detection, autonomous decision-making, and institutional governance, these outputs are often treated as indicators of machine understanding. This talk argues that such interpretation reflects a cognitive illusion rather than a technological achievement. Large-scale AI models generate statistically coherent language and structured responses, but they do not possess intention, awareness, or interpretive continuity. The appearance of understanding emerges from fluency and pattern recognition, not from cognition. When fluency is mistaken for comprehension, institutions begin to over-attribute agency and epistemic reliability to systems that do not understand their own outputs. In low-risk contexts, this illusion may remain manageable. In high-stakes environments, however, the misrecognition becomes consequential. Decision-makers may over-trust generated explanations, treat simulated reasoning as genuine analysis, and shift responsibility onto automated systems while human oversight becomes reactive rather than reflective. This presentation examines the psychological and institutional mechanisms that lead to the attribution of understanding to non-understanding systems. It explores how simulated cognition reshapes human judgment, redistributes cognitive labor, and alters responsibility structures in AI-mediated environments. Rather than focusing on technical robustness alone, the talk proposes that sustainable AI integration requires epistemic clarity: recognizing the difference between fluent simulation and genuine understanding. The central claim is simple but urgent: advanced AI systems generate the appearance of thought, but the responsibility for meaning, judgment, and accountability remains entirely human.