Dr. Auday Berro is an Assistant Professor at ESILV, where he leads the Generative AI and Agentic AI minor. He is a member of the STARCS research axis. He received his PhD in Computer Science from Université Claude Bernard Lyon 1, France, in June 2024. He has also taught at École Centrale de Lyon and IUT Lyon 1. His doctoral research explored how paraphrase generation can make human-chatbot conversations more natural and less frustrating (improving lexical and syntactic diversity), guided by the principle "Chatbots for machines, conversations for humans". During his postdoctoral research, he investigated AI-driven digital health, focusing on how reinforcement learning and interoperability can support personalized, adaptive interventions to improve sleep and mental well-being. From making chatbot conversations more diverse to turning data into better sleep quality, his research explored different ways to use AI to enhance the human experience?and, over time, led to the publication of various scientific papers. His current research focuses on Human-Centered Artificial Intelligence (HCAI), particularly the evaluation of reasoning in large language models (LLMs). He investigates what it means for an LLM to reason and how reasoning capabilities can be rigorously assessed. His work adopts an interdisciplinary approach, drawing on neuroscience, neurophysiology, cognitive science and learning sciences to investigate how reasoning is conceptualized, studied and evaluated across different fields. Through a critical evaluation of the theoretical assumptions, evaluation methods and empirical data behind current approaches to reasoning in natural language processing (NLP), AI and research on LLMs, his work aims to bridge the gap between current practices in the field of computer science and scientific knowledge about human cognition and intelligence developed in fields such as neuroscience, cognitive science and learning sciences. This perspective aims to contribute to the development of more rigorous, theory-based, and human-centered approaches to understanding and evaluating AI systems.
No posts by this author.
