Addison J. Wu

Addison J. Wu

addisonwu@princeton.edu | X | Google Scholar | LinkedIn | GitHub | CV

I'm a rising senior at Princeton University studying computer science and math, where I'm advised by Tom Griffiths and Zhuang Liu as a part of Princeton Language and Intelligence. Currently, I'm also a research intern at the Center for AI Safety, advised by Mantas Mazeika and Dan Hendrycks. I got my first taste of research at the Multi-Physics Interaction Lab, University of Waterloo, advised by Jean-Pierre Hickey

I want AI systems to interact better with each other, with us, and with the world over time so we can do even greater things we find worth doing

Novel social biases
Large Language Models Develop Novel Social Biases Through Adaptive Exploration
Addison J. Wu*, Ryan Liu*, Xuechunzi Bai, Thomas L. Griffiths
ICML 2026 Oral [arXiv]
Press coverage: The Guardian, Forbes, MIT Technology Review, The Times of India, Gizmodo, Arizona's Family
Ads in AI chatbots
Ads in AI Chatbots? An Analysis of How Large Language Models Navigate Conflicts of Interest
Addison J. Wu*, Ryan Liu*, Shuyue Stella Li, Yulia Tsvetkov, Thomas L. Griffiths
COLM 2026 [arXiv]
Press coverage: Hacks/Hackers, The Indian Express; Cited in Senate testimony
Polysemous words in multimodal models
Where did the ambiguity go? Examining how multimodal models interpret polysemous words
Jasin Cekinmez*, Addison J. Wu*, Raja Marjieh, Thomas L. Griffiths
Sci-FM @ COLM 2026 Oral [arXiv]
SportD
SportD: Can VLMs Physically Strategize?
Jasin Cekinmez*, Addison J. Wu*, Haotian Xia, Akshaya Bharadhwaj, Anay Putty, Anirudh Ravishankar, Jaewoong Lee, Jinglin Xiao, Kyumin Andrew Shim, Mishika Ahuja, Nisarga Patil, Leo Liu, Zhuohan Liu, Weining Shen
arXiv preprint, 2026 [arXiv]
Compute-supervision tradeoffs in RLVR
Quantifying Empirical Compute-Supervision Tradeoffs in RLVR
Ryo Mitsuhashi*, Patrick Chen*, Isabelle Tseng*, Jasin Cekinmez*, Addison J. Wu*
Workshop on Combining Benchmarks and Theory @ ICML 2026 [arXiv]
OccludeBench
OccludeBench: Can VLMs See the Hint?
Jasin Cekinmez*, Addison J. Wu*, Ryo Mitsuhashi*, Linrong Cai, Xingyu Fu, Zhuang Liu
Computer Vision in the Wild @ CVPR 2026
Guess the unified model
Guess the Unified Model: How Much Can We Recover from Generated Images?
Jasin Cekinmez*, Ryo Mitsuhashi*, Addison J. Wu*, Yida Yin
under review, 2026 [arXiv]
Query timing and positional bias
Query Timing Produces Opposite Positional Biases Between LLMs and Humans
Jasin Cekinmez*, Addison J. Wu*, Thomas L. Griffiths
ICBINB @ ICLR 2026 Entropic Paper Award [Paper]
Motives behind communication
Are Large Language Models Sensitive to the Motives Behind Communication?
Addison J. Wu*, Ryan Liu*, Kerem Oktar*, Theodore R. Sumers, Thomas L. Griffiths
NeurIPS 2025; PragLM @ COLM 2025 Oral [arXiv]
Mind your step (by step)
Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse
Ryan Liu*, Jiayi Geng*, Addison J. Wu, Ilia Sucholutsky, Tania Lombrozo, Thomas L. Griffiths
ICML 2025 [arXiv]
Press coverage: The Information [text]
Infrasound source localization
Toward a Neural Network-Based Approach for Improved Atmospheric Infrasound Source Localization
Addison J. Wu, Arnav Joshi, Jean-Pierre Hickey
IEEE Access, 2025 [Paper]

I love to travel ✈️
🇨🇦 🇺🇸 🇲🇽 🇵🇷 🇵🇦 🇧🇷 🇯🇵 🇰🇷 🇨🇳 🇭🇰 🇮🇪 🇬🇧 🇫🇷 🇪🇸 🇧🇪 🇨🇭 🇮🇹 🇻🇦 🇸🇲 🇩🇰 🇸🇪 🇳🇴