The Future is Agentic in Recommender Systems

Kyle Polich sits down with Yashar Deldjoo, Associate Professor at the Polytechnic University of Bari, to explore how recommender systems have evolved and why trustworthiness matters. They unpack key dimensions of responsible AI, including robustness to adversarial attacks, privacy, explainability, and fairness, and discuss how LLMs introduce new risks like hallucinations. The episode closes with a look at “agentic” recommender systems, where tools and memory shift recommendations from ranked lists to end-to-end task completion.

Guest

Yashar Deldjoo: Dr. Yashar Deldjoo is a senior research scientist and Associate Professor of Computer Science at the Polytechnic University of Bari, Italy. His research agenda focuses on integrating trustworthy and responsible AI into modern recommender systems, particularly those leveraging Generative AI and multi‑agent large language models. Over the past several years, he has played a pivotal role in understanding, categorizing and mitigating emerging risks in these systems—including hallucinations, biases, reasoning challenges, content‑safety concerns, adversarial resilience and goal misalignment—and in developing holistic evaluation frameworks to ensure safe deployment. Dr. Deldjoo recently led a multidisciplinary collaboration to author Next‑Generation Recommender Systems under Generative AI (Gen‑RecSys), a forthcoming book within the Foundations and Trends® in Information Retrieval series (FNT&IR 2025), which explores advanced generative techniques for recommender systems. Beyond research, Dr. Deldjoo provides extensive service to the academic community. He is an Associate Editor for IEEE Transactions on Knowledge and Data Engineering (TKDE), ACM Computing Surveys (CSUR), and ACM Transaction on Recommender Systems (TORS). He has served as a Track Chair or Senior Program Committee member at major IR and AI conferences, including KDD, The Web Conference (WWW), CIKM, SIGIR, RecSys, and ECAI, among others.