The Future is Agentic in Recommender Systems

Recommendation systems quietly shape what we watch, read, buy, and discover.

In this episode, host Kyle Polich talks with Yashar Deldjoo, research scientist and Associate Professor at the Polytechnic University of Bari and a senior research scientist with a decade of experience in recommender systems. Yashar traces their path from early work in image processing and multimedia to a PhD focused on video recommendation, and then into a research agenda centered on trustworthy and responsible recommender systems. The conversation breaks down what “trustworthy AI” means in practice for recommenders, including core dimensions such as generalizability, robustness to adversarial manipulation, privacy, explainability, and fairness. Yashar explains how large language models are reshaping recommender systems by enabling conversational interaction and moving the goal from “ranking items” to “doing tasks” under multiple real-world constraints. The episode also covers emerging risks such as hallucinations and safety concerns, plus how agentic systems, tools, and different kinds of memory (short-term, episodic, semantic, and procedural) may shape the next generation of recommendations.

Yashar Deldjoo

Polytechnic University of Bari

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