I am a PhD student in statistics at the University of Toronto, supervised by Prof. Zhijing Jin and Prof. Dehan Kong. My research focuses on the theory of causal inference and causal inference for large language models (LLMs).
I currently work on causal decision-making and causal inference with unstructured data.
Contact: yahang.qi@mail.utoronto.ca
My research goal is to use causal inference to understand the causal mechanisms of complex systems and improve decisions within those systems. I develop methods that use large language models (LLMs) to bring expert knowledge into the causal inference process, and I also use causal inference to understand LLMs and improve their performance.
I develop methods that use LLMs as efficient tools for bringing external knowledge into practical causal inference. This helps build causal inference methods that can work under more relaxed assumptions.
I develop theory and computational methods that use causal models as causal abstractions of LLM behavior. This makes it possible to study LLMs through simpler causal representations of otherwise complex and computationally costly systems.
Causal inference describes mechanisms in complex systems, making it a useful tool for science in the agentic AI regime. I study how causal methods can identify candidate causal relationships and propose valuable hypotheses.