About
I am a fourth-year PhD student at MIT BCS and CSAIL, advised by Josh Tenenbaum and Roger Levy. Previously, I completed my undergraduate degree at UC Berkeley with backgrounds in cognitive science, computer science, logic, and philosophy.
My full name is Cedegao Zhang, and I go by Ced.
Email: cedzhang [at] mit [dot] edu
Research overview
I'm interested in the following areas from interdisciplinary perspectives:
- Artificial intelligence
- Reasoning, broadly defined
- Natural language understanding and generation
- Language models, agents, and alignment
- Computational cognitive science
- Learning, thinking, and problem solving
- Communication and social cognition
- Philosophy
- of language, mind, science, and AI
Questions I like to think about these days: How should we evaluate and improve language models and associated agents/systems? What are good approaches to combine LMs with rational computations (programs, Bayes)? In what sense are humans better learners, reasoners, and communicators than state-of-the-art AI? And how can we make AI more data-efficient, reliable, and aligned? How should we think about the relationships and interactions between humans and AI, both individually and collectively?
On top of these interests and questions, my long-term goal is to (1) build general AI agents in a human-inspired and human-compatible way with a focus on language, (2) facilitate human-AI symbiosis, and (3) advance the understanding of intelligence through theories and models.