High-stakes decisions under uncertainty, such as medical emergency triage, require more than accurate predictions. They depend on estimating the likelihood of alternative outcomes while explicitly weighing the consequences of different actions, principles that have long formed the foundation of medical diagnosis and de...
K. Yamin, Christopher Kelly, Bryan Wilder et al.· 1 citation
Inferring intelligence from observable behavior is a foundational challenge in artificial intelligence. We develop a theory of Bayesian intelligence for agents such as language models. Each prompt induces a possibly imperfect internal experiment; the agent updates a full-support prior by Bayes'rule and faithfully repor...
An empirical pipeline is presented for estimating the implied preferences that an LLM's observed choices optimize: the model's probability distribution over unknowns is elicited along with the choice it would make for the decision task and a discrete choice model is fit to recover the cost function that best rationaliz...
K. Yamin, Jingjing Tang, Eric Horvitz et al.· arXiv.org· 4 citations
There is significant uncertainty about whether abstractions like beliefs or desires usefully describe the behavior of large language models (LLMs). In addition to the inherent scientific interest of this question, these latent quantities are often invoked to explain the behavior of LLMs to users or to define and evalua...
Alex Smolin, Bryan Wilder· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
