Beyond demographics: aligning role-playing LLM-based agents using human belief networks
Creating human-like large language model (LLM) agents is crucial for faithful social
simulation. Having LLMs role-play based on demographic information sometimes improves …
simulation. Having LLMs role-play based on demographic information sometimes improves …
Dimensions of disagreement: Divergence and misalignment in cognitive science and artificial intelligence.
Our understanding of disagreement is rooted in psychological studies of human behavior,
which typically cast disagreement as divergence: two agents forming diverging evaluations …
which typically cast disagreement as divergence: two agents forming diverging evaluations …
Pipeline for modeling causal beliefs from natural language
We present a causal language analysis pipeline that leverages a Large Language Model to
identify causal claims made in natural language documents, and aggregates claims across …
identify causal claims made in natural language documents, and aggregates claims across …
How rational inference about authority debunking can curtail, sustain, or spread belief polarization
In polarized societies, divided subgroups of people have different perspectives on a range of
topics. Aiming to reduce polarization, authorities may use debunking to lend support to one …
topics. Aiming to reduce polarization, authorities may use debunking to lend support to one …
Learning from and about climate scientists
Despite the overwhelming scientific consensus that human activities contribute significantly
to climate change, public opinion remains divided. To bridge this gap, informative …
to climate change, public opinion remains divided. To bridge this gap, informative …
How aggregated opinions shape beliefs
K Oktar, T Lombrozo - Nature Reviews Psychology, 2025 - nature.com
In today's online world, the beliefs of people are shaped by aggregated opinions: the
elicited, quantified and summarized judgements of many strangers. Ratings guide …
elicited, quantified and summarized judgements of many strangers. Ratings guide …
A Bayesian decision-theoretic framework for studying motivated reasoning
Psychological, political, cultural, and sociological factors shape how people form and revise
their beliefs. An established finding across these fields is that people are motivated to hold …
their beliefs. An established finding across these fields is that people are motivated to hold …
TAXI: Evaluating Categorical Knowledge Editing for Language Models
Humans rarely learn one fact in isolation. Instead, learning a new fact induces knowledge of
other facts about the world. For example, in learning a korat is a type of cat, you also infer it …
other facts about the world. For example, in learning a korat is a type of cat, you also infer it …