Human-Centric AI with Common Sense by Filip Ilievski

Human-Centric AI with Common Sense by Filip Ilievski

This book enables readers to understand the challenges and opportunities of developing human-centered AI with commonsense reasoning abilities. Despite apparent accuracy improvements brought by large neural models across task benchmarks, common sense is still lacking. The lack of common sense affects many tasks, including story understanding, decision-making, and question answering. Commonsense knowledge and reasoning have long been considered the “black matter” of AI, raising concerns about the trustworthiness and applicability of AI methods in both autonomous and hybrid applications. This book describes how to design a more robust, collaborative, explainable, and responsible AI through incorporating neuro-symbolic commonsense reasoning. In addition, the book provides examples of how these properties of AI can facilitate a wide range of social-good applications in digital democracy, traffic monitoring, education, and robotics. What makes commonsense reasoning such a unique and impactful challenge? What can we learn from cognitive research when designing and developing AI systems? How can we approach building responsible, robust, collaborative, and explainable AI with common sense? And finally, what is the impact of this work on human-AI teaming? This book provides an accessible introduction and exploration of these topics.

Filip Ilievski, Ph.D., is a Senior Assistant Professor of Commonsense AI at the VrijeUniversiteit (VU) Amsterdam’s Computer Science department. Dr. Ilievski is an affiliated scientist at the USC Information Sciences Institute (ISI), where he was previously a Research Lead and played a key role in a team within the DARPAMachine Common Sense (MCS) program. Dr. Ilievski holds a Ph.D. in Natural Language Processing. His research focuses on developing robust and explainable neuro-symbolic technology with positive real-world impact, based on neural methods and high-quality knowledge. Dr. Ilievski has also made extensive contributions in identifying long-tail entities in text, interpretation of internet memes, and enabling access to large-scale knowledge resources.