Language Models Should be Used to Surface the Unwritten Code of Science and Society

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Bao, Honglin, Wu, Siyang, Choi, Jiwoong, Mao, Yingrong, Evans, James A.
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918306688008192
author Bao, Honglin
Wu, Siyang
Choi, Jiwoong
Mao, Yingrong
Evans, James A.
author_facet Bao, Honglin
Wu, Siyang
Choi, Jiwoong
Mao, Yingrong
Evans, James A.
contents This paper calls on the research community not only to investigate how human biases are inherited by large language models (LLMs) but also to explore how these biases in LLMs can be leveraged to make society's "unwritten code" - such as implicit stereotypes and heuristics - visible and accessible for critique. We introduce a conceptual framework through a case study in science: uncovering hidden rules in peer review - the factors that reviewers care about but rarely state explicitly due to normative scientific expectations. The idea of the framework is to push LLMs to speak out their heuristics through generating self-consistent hypotheses - why one paper appeared stronger in reviewer scoring - among paired papers submitted to 46 academic conferences, while iteratively searching deeper hypotheses from remaining pairs where existing hypotheses cannot explain. We observed that LLMs' normative priors about the internal characteristics of good science extracted from their self-talk, e.g., theoretical rigor, were systematically updated toward posteriors that emphasize storytelling about external connections, such as how the work is positioned and connected within and across literatures. Human reviewers tend to explicitly reward aspects that moderately align with LLMs' normative priors (correlation = 0.49) but avoid articulating contextualization and storytelling posteriors in their review comments (correlation = -0.14), despite giving implicit reward to them with positive scores. These patterns are robust across different models and out-of-sample judgments. We discuss the broad applicability of our proposed framework, leveraging LLMs as diagnostic tools to amplify and surface the tacit codes underlying human society, enabling public discussion of revealed values and more precisely targeted responsible AI.
format Preprint
id arxiv_https___arxiv_org_abs_2505_18942
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Language Models Should be Used to Surface the Unwritten Code of Science and Society
Bao, Honglin
Wu, Siyang
Choi, Jiwoong
Mao, Yingrong
Evans, James A.
Computers and Society
Computation and Language
Digital Libraries
This paper calls on the research community not only to investigate how human biases are inherited by large language models (LLMs) but also to explore how these biases in LLMs can be leveraged to make society's "unwritten code" - such as implicit stereotypes and heuristics - visible and accessible for critique. We introduce a conceptual framework through a case study in science: uncovering hidden rules in peer review - the factors that reviewers care about but rarely state explicitly due to normative scientific expectations. The idea of the framework is to push LLMs to speak out their heuristics through generating self-consistent hypotheses - why one paper appeared stronger in reviewer scoring - among paired papers submitted to 46 academic conferences, while iteratively searching deeper hypotheses from remaining pairs where existing hypotheses cannot explain. We observed that LLMs' normative priors about the internal characteristics of good science extracted from their self-talk, e.g., theoretical rigor, were systematically updated toward posteriors that emphasize storytelling about external connections, such as how the work is positioned and connected within and across literatures. Human reviewers tend to explicitly reward aspects that moderately align with LLMs' normative priors (correlation = 0.49) but avoid articulating contextualization and storytelling posteriors in their review comments (correlation = -0.14), despite giving implicit reward to them with positive scores. These patterns are robust across different models and out-of-sample judgments. We discuss the broad applicability of our proposed framework, leveraging LLMs as diagnostic tools to amplify and surface the tacit codes underlying human society, enabling public discussion of revealed values and more precisely targeted responsible AI.
title Language Models Should be Used to Surface the Unwritten Code of Science and Society
topic Computers and Society
Computation and Language
Digital Libraries
url https://arxiv.org/abs/2505.18942