Presented by Dr Taotao Cai
|
From Agent Traces to Trustworthy Memory: Provenance, Governance, and Repair in LLM Agents
Large language model agents are evolving from passive text generators into stateful systems that retrieve information, use tools, maintain memory, and take actions. This talk will discuss how provenance and evidence tracing can support trustworthy agent memory, runtime governance, and recovery from incorrect or outdated beliefs, and will highlight several open challenges in building traceable, governable, and recoverable AI agents.
Please join us for this upcoming seminar featuring Dr Taotao Cai as our respected speaker.
Dr Taotao Cai is a Senior Lecturer in Computing at the University of Southern Queensland (UniSQ). His research spans graph learning, trustworthy AI, and self-improving LLM agents, with current interests in provenance, reliability, and AI for networked and industrial decision-making. He has published more than 30 peer-reviewed papers in venues including IEEE TKDE, ICDE, IJCAI, IEEE TGRS, and IEEE Transactions on Services Computing.
Since 2025, he has contributed to more than A$1.6 million in competitive and industry-linked research projects. He is Lead Chief Investigator of the A$549,886 AEA Ignite SafeBuild AI project, a Chief Investigator on the A$508,442 AEA Ignite Intelligent FCAS Agent project, and Lead Supervisor of one of two Australian Government National Industry PhD projects undertaken with industry partners.
Please share amongst your networks and if you have any questions, please reach out to the CSAS ROA, via email at CSAS@unisq.edu.au or Professor Tek Maraseni (CSAS Deputy Director and Seminar Moderator) on Tek.Maraseni@unisq.edu.au.
Kind regards, Centre for Sustainable Agricultural Systems |
