Squirrel AI Co-founder Wen Qingsong Delivers Keynote at IEEE SWC 2026 on Scaling Agentic AI Tutors
In October 2026, the 12th IEEE International Smart World Congress (IEEE SWC 2026) was held in Lamezia Terme, Calabria, Italy. Under the theme “Convergence of Minds: Forging Human-AI Partnerships Across Disciplines,” the congress highlighted generative AI-driven education as a key topic.
Dr. Wen Qingsong, co-founder and chief scientist of Squirrel AI, delivered a keynote titled “Agentic AI Tutors for Smart Education at Scale.” The talk explored how agentic AI tutors can enable adaptive learning and personalized instruction in real educational settings while moving toward large-scale deployment.
Wen explained that agentic AI tutors extend AI capabilities to multi-agent large model reasoning, multimodal error analysis, dialogue-driven interaction, and long-term learner modeling. These systems can perform fine-grained diagnosis of knowledge gaps, generate personalized feedback, and optimize adaptive learning paths. Unlike traditional question-answering AI, agentic tutors must continuously understand a student’s state and adjust teaching actions based on new information.
The keynote emphasized that scaling AI education requires maintaining personalized pacing and learning paths even as the number of learners grows. Continuous learner profiling is foundational, with Squirrel AI’s multimodal adaptive education large model integrating learning behavior data, subject-specific fine-grained knowledge graphs, and adaptive capabilities.
Wen also stressed that reliability, robustness, and trustworthiness are critical technical challenges for agentic AI tutors entering real educational environments. Model capability determines what AI can do, while these three dimensions determine whether AI can sustain long-term teaching roles.
