Research on large language models, explained.
I’m Fan Huang, a Ph.D. candidate in Informatics at Indiana University Bloomington. I study human–AI alignment: probing and mitigating the biases and limitations of large language models. Here I share my ongoing work, with articles that explain what they find in plain language.
Latest
Playing Civilization VI with AI agents: learning, adopting and creating strategy
A primer on how AI agents can play a strategy game like Civilization VI: how they learn strategies, put them to use and invent new ones, what happened when frontier models played full games, and what still stands in the way.
What does a language model believe? It depends on how you ask
A plain-language guide to the LLM Belief Atlas, and to why a model’s stance is not a single number.
The LLM Belief Atlas: Visualizing How Language-Model Stances Depend on Elicitation
What a language model appears to believe about a proposition is not a scalar but a signal that depends on how it is elicited and read: forcing a choice can fabricate a value, option-logprob scoring carries order bias, and sycophancy makes…