Beyond LLMs: Can we build AI systems that not only generate answers, but also understand and explain them?
At the inauguration of Artificial Intelligence Research Institute - AIRi@UTCN AIRi@UTCN – Artificial Intelligence Research Institute, during AI Forum 2026, we had the pleasure of welcoming Prof. Mihai Surdeanu from the University of Arizona, who delivered an insightful presentation on “Beyond LLMs: Neuro-Symbolic Approaches for Language Understanding.” Large Language Models have achieved remarkable success. Yet, as Prof. Surdeanu highlighted, important challenges remain: hallucinations, limited explainability, and difficulties in controlling and maintaining AI systems as requirements evolve. These limitations become particularly critical in high-stakes domains such as medicine, where a plausible answer is not enough. We need AI systems that can reason reliably and explain their conclusions. Prof. Surdeanu presented a promising direction: Neuro-Symbolic AI, combining the best of two worlds: 🧠 Neural AI — the ability to learn patterns, generalize, and process complex information. 🔗 Symbolic AI — the ability to represent knowledge explicitly, structure reasoning, and provide interpretable explanations. One particularly compelling example came from biomedical applications, where symbolic knowledge guides LLM reasoning by identifying the contextual information most relevant to understanding causal mechanisms. The result? In some cases, a twofold improvement in LLM performance. This presentation strongly resonates with our vision for AIRi@UTCN: moving beyond AI systems that merely generate convincing answers toward systems that can reason, explain, and earn our trust. It also reinforces an important research direction for our institute: bringing together machine learning, knowledge representation, reasoning, and explainable AI to address real-world challenges. We sincerely thank Prof. Mihai Surdeanu for contributing to the scientific program of AI Forum 2026 and for sharing research that challenges us to think beyond the current LLM paradigm. Prof. Mihai Mihai Surdeanu is a Professor of Computer Science at the University of Arizona, with appointments in Linguistics and Cognitive Science. His research focuses on natural language understanding, information extraction, and interpretable AI. He previously conducted research at Stanford's NLP group and served as Chief Scientist at Lex Machina. The future of AI is not just about larger models. It is about better reasoning, deeper understanding, and greater trust. #AIRiUTCN #AIForum2026 #NeuroSymbolicAI #ExplainableAI #TrustworthyAI #LLM #KnowledgeRepresentation #ArtificialIntelligence #UniversityOfArizona
