@inproceedings{andreas-2022-language,
title = "Language Models as Agent Models",
author = "Andreas, Jacob",
editor = "Goldberg, Yoav and
Kozareva, Zornitsa and
Zhang, Yue",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022",
month = dec,
year = "2022",
address = "Abu Dhabi, United Arab Emirates",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.findings-emnlp.423",
doi = "10.18653/v1/2022.findings-emnlp.423",
pages = "5769--5779",
abstract = "Language models (LMs) are trained on collections of documents, written by individual human agents to achieve specific goals in the outside world. During training, LMs have access only to text of these documents, with no direct evidence of the internal states of the agents that produced them{---}a fact often used to argue that LMs are incapable of modeling goal-directed aspects of human language production and comprehension. Can LMs trained on text learn anything at all about the relationship between language and use? I argue that LMs are models of communicative intentions in a specific, narrow sense. When performing next word prediction given a textual context, an LM can infer and represent properties of an agent likely to have produced that context. These representations can in turn influence subsequent LM generation in the same way that agents{'} communicative intentions influence their language. I survey findings from the recent literature showing that{---}even in today{'}s non-robust and error-prone models{---}LMs infer and use representations of fine-grained communicative intentions and high-level beliefs and goals. Despite the limited nature of their training data, they can thus serve as building blocks for systems that communicate and act intentionally.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="andreas-2022-language">
<titleInfo>
<title>Language Models as Agent Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jacob</namePart>
<namePart type="family">Andreas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2022</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yoav</namePart>
<namePart type="family">Goldberg</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zornitsa</namePart>
<namePart type="family">Kozareva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Abu Dhabi, United Arab Emirates</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Language models (LMs) are trained on collections of documents, written by individual human agents to achieve specific goals in the outside world. During training, LMs have access only to text of these documents, with no direct evidence of the internal states of the agents that produced them—a fact often used to argue that LMs are incapable of modeling goal-directed aspects of human language production and comprehension. Can LMs trained on text learn anything at all about the relationship between language and use? I argue that LMs are models of communicative intentions in a specific, narrow sense. When performing next word prediction given a textual context, an LM can infer and represent properties of an agent likely to have produced that context. These representations can in turn influence subsequent LM generation in the same way that agents’ communicative intentions influence their language. I survey findings from the recent literature showing that—even in today’s non-robust and error-prone models—LMs infer and use representations of fine-grained communicative intentions and high-level beliefs and goals. Despite the limited nature of their training data, they can thus serve as building blocks for systems that communicate and act intentionally.</abstract>
<identifier type="citekey">andreas-2022-language</identifier>
<identifier type="doi">10.18653/v1/2022.findings-emnlp.423</identifier>
<location>
<url>https://aclanthology.org/2022.findings-emnlp.423</url>
</location>
<part>
<date>2022-12</date>
<extent unit="page">
<start>5769</start>
<end>5779</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Language Models as Agent Models
%A Andreas, Jacob
%Y Goldberg, Yoav
%Y Kozareva, Zornitsa
%Y Zhang, Yue
%S Findings of the Association for Computational Linguistics: EMNLP 2022
%D 2022
%8 December
%I Association for Computational Linguistics
%C Abu Dhabi, United Arab Emirates
%F andreas-2022-language
%X Language models (LMs) are trained on collections of documents, written by individual human agents to achieve specific goals in the outside world. During training, LMs have access only to text of these documents, with no direct evidence of the internal states of the agents that produced them—a fact often used to argue that LMs are incapable of modeling goal-directed aspects of human language production and comprehension. Can LMs trained on text learn anything at all about the relationship between language and use? I argue that LMs are models of communicative intentions in a specific, narrow sense. When performing next word prediction given a textual context, an LM can infer and represent properties of an agent likely to have produced that context. These representations can in turn influence subsequent LM generation in the same way that agents’ communicative intentions influence their language. I survey findings from the recent literature showing that—even in today’s non-robust and error-prone models—LMs infer and use representations of fine-grained communicative intentions and high-level beliefs and goals. Despite the limited nature of their training data, they can thus serve as building blocks for systems that communicate and act intentionally.
%R 10.18653/v1/2022.findings-emnlp.423
%U https://aclanthology.org/2022.findings-emnlp.423
%U https://doi.org/10.18653/v1/2022.findings-emnlp.423
%P 5769-5779
Markdown (Informal)
[Language Models as Agent Models](https://aclanthology.org/2022.findings-emnlp.423) (Andreas, Findings 2022)
ACL
- Jacob Andreas. 2022. Language Models as Agent Models. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 5769–5779, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.