Speaker
Description
This talk provides an overview of recent advances in agentic AI and deep learning at BESIII. I will introduce DrSai, a multi-agent system designed for autonomous end-to-end physics analysis—from natural-language goals to data selection, statistical inference, and systematic evaluation—with emphasis on reliability, traceability, and human-AI collaboration. I will also overview our deep-learning efforts, including an experimental-data foundation model, SAM-based reconstruction of intermediate resonances, and machine-learning-based beam-background suppression in the BESIII trigger. Finally, I will discuss the evolving role of AI—from an efficiency tool to an active reasoning partner in scientific discovery—and how human-AI collaboration can accelerate high-energy physics and beyond.
