Date: Tuesday, September 8
Time: 09.00
Venue:https://meet.google.com/spv-duqx-aeh 
Speaker: Indro Spinelli

Title: From Trustworthy Models to Trustworthy Agent Societies

Abstract: As AI evolves from isolated models into autonomous agents interacting with humans, robots, and one another, trustworthiness must become an intrinsic property rather than an external safeguard. In this talk, we trace our research trajectory toward Trustworthy Agentic Intelligence, spanning fair and explainable learning, human-motion generation, machine unlearning, and safety-aware robotics. We then turn to the emerging challenge of multi-agent societies, examining how agent identity can distort coordination and how efficient architectures can make capable agents more affordable to deploy. This leads to a broader question: how can societies of AI agents scale in capability and complexity while remaining reliable, interpretable, and efficient?