Perceptions of Linguistic Uncertainty by Language Models and Humans

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Belem, Catarina G, Kelly, Markelle, Steyvers, Mark, Singh, Sameer, Smyth, Padhraic
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910687520882688
author Belem, Catarina G
Kelly, Markelle
Steyvers, Mark
Singh, Sameer
Smyth, Padhraic
author_facet Belem, Catarina G
Kelly, Markelle
Steyvers, Mark
Singh, Sameer
Smyth, Padhraic
contents _Uncertainty expressions_ such as "probably" or "highly unlikely" are pervasive in human language. While prior work has established that there is population-level agreement in terms of how humans quantitatively interpret these expressions, there has been little inquiry into the abilities of language models in the same context. In this paper, we investigate how language models map linguistic expressions of uncertainty to numerical responses. Our approach assesses whether language models can employ theory of mind in this setting: understanding the uncertainty of another agent about a particular statement, independently of the model's own certainty about that statement. We find that 7 out of 10 models are able to map uncertainty expressions to probabilistic responses in a human-like manner. However, we observe systematically different behavior depending on whether a statement is actually true or false. This sensitivity indicates that language models are substantially more susceptible to bias based on their prior knowledge (as compared to humans). These findings raise important questions and have broad implications for human-AI and AI-AI communication.
format Preprint
id arxiv_https___arxiv_org_abs_2407_15814
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Perceptions of Linguistic Uncertainty by Language Models and Humans
Belem, Catarina G
Kelly, Markelle
Steyvers, Mark
Singh, Sameer
Smyth, Padhraic
Computation and Language
Artificial Intelligence
Machine Learning
_Uncertainty expressions_ such as "probably" or "highly unlikely" are pervasive in human language. While prior work has established that there is population-level agreement in terms of how humans quantitatively interpret these expressions, there has been little inquiry into the abilities of language models in the same context. In this paper, we investigate how language models map linguistic expressions of uncertainty to numerical responses. Our approach assesses whether language models can employ theory of mind in this setting: understanding the uncertainty of another agent about a particular statement, independently of the model's own certainty about that statement. We find that 7 out of 10 models are able to map uncertainty expressions to probabilistic responses in a human-like manner. However, we observe systematically different behavior depending on whether a statement is actually true or false. This sensitivity indicates that language models are substantially more susceptible to bias based on their prior knowledge (as compared to humans). These findings raise important questions and have broad implications for human-AI and AI-AI communication.
title Perceptions of Linguistic Uncertainty by Language Models and Humans
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2407.15814