Wentao Wan

Welcome to my homepage! I recently earned my Ph.D. in Artificial Intelligence (Dec. 2025) from the School of Computer Science and Engineering at Sun Yat-sen University. I was a member of the HCP Lab, fortunately advised by Prof. Liang Lin and Prof. Keze Wang.

Research Statement:

I believe that intelligence is a natural phenomenon in our universe. Driven by a deep curiosity about this phenomenon, I chose Artificial Intelligence as my research field, aiming to achieve a fundamental understanding of intelligence through the very act of creating it. I have a broad interest in various phenomena of intelligence.

During my Ph.D., my research interests primarily focused on Visual Understanding, LLM-based Agents, Rigorous Reasoning of LLMs, and Continual Learning (please refer to my CV for details). My doctoral dissertation, titled “Research on Enhancing Cognition and Behavior of Large Language Model-based Agents”, developed methods to bolster the general reasoning and knowledge-updating capacities of LLMs——the cognitive core of agents——and addressed efficient tool learning within non-differentiable agent systems for Visual Reasoning.

Through my doctoral research and reflections, I have developed a personal definition of intelligence that differs from the Turing Test: an adaptive computational process for understanding the world or changing it to achieve specific goals.

Guided by this philosophy, my current research interests focus on the following three directions (I will further elaborate on these in my blog posts):

• World Models & Physical Intelligence: Developing next-generation models and learning paradigms capable of truly understanding and purposefully reshaping the real world.
• Data-Efficient Learning: Exploring highly data-efficient (reinforcement) learning paradigms for the era of experiential learning.
• Authentic Continual Learning: Establishing robust continual learning paradigms for the long-term, open-ended evolution of agents.

Ultimately, I aim to accumulate extensive practical experience and theoretical intuition during this journey, enabling me to achieve a deeper theoretical understanding of intelligence. In turn, I hope to use this theoretical understanding to better guide the practice of building intelligent systems.

I am currently on the job market and actively seeking research positions dedicated to one or more of the above directions. Any interested collaborators are warmly welcome to contact me via email (wanwentao93@gmail.com).