Epistemological challenges of AI | IASEAI '26

What does it mean for an AI system to “know” something—and how can we tell when that knowledge is reliable? This breakout session from IASEAI 2026 examines how AI systems challenge our understanding of knowledge, reasoning, and truth. From ideological bias and model behavior to questions of consciousness and disciplinary fragmentation, speakers explore how we interpret — and often misinterpret — what AI systems “know.” Across two segments, the session interrogates the limits of evaluation, the risks of anthropomorphizing AI, and the growing gap between technical development and broader societal understanding. Tijl De Bie – Large Language Models Reflect the Ideology of their Creators Jessica Bo – Language Models Exhibit Inconsistent Biases Towards Algorithmic Agents and Human Experts Anna Neumann – AI Safety Evaluations Need To Consider Cascading Effects Sahil Kale – Mirage of Mastery: Memorization Tricks LLMs into Artificially Inflated Self-Knowledge Syed AbuMusab – Blaming AI is Pointless: Against Using Social and Moral Concepts to Explain AI Behavior Oisín Hugh Clancy – Attribution and Actualisation of Consciousness in AI Dror Markus – Societal AI Research Has Become Less Interdisciplinary Dani Roytburg – Mind the Gap! Pathways Towards Unifying AI Safety and Ethics Research 📍 Recorded at UNESCO Headquarters, Paris The International Association for Safe and Ethical AI (IASEAI) is a global professional association bringing together researchers, policymakers, and practitioners working to ensure that advanced AI systems operate safely and ethically for the benefit of humanity. Learn more: https://www.iaseai.org #AI #AISafety #AIGovernance #AIpolicy #IASEAI #ArtificialIntelligence #ResponsibleAI