Examining Identity Drift in Conversations of LLM Agents

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Choi, Junhyuk, Hong, Yeseon, Kim, Minju, Kim, Bugeun
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929716734197760
author Choi, Junhyuk
Hong, Yeseon
Kim, Minju
Kim, Bugeun
author_facet Choi, Junhyuk
Hong, Yeseon
Kim, Minju
Kim, Bugeun
contents Large Language Models (LLMs) show impressive conversational abilities but sometimes show identity drift problems, where their interaction patterns or styles change over time. As the problem has not been thoroughly examined yet, this study examines identity consistency across nine LLMs. Specifically, we (1) investigate whether LLMs could maintain consistent patterns (or identity) and (2) analyze the effect of the model family, parameter sizes, and provided persona types. Our experiments involve multi-turn conversations on personal themes, analyzed in qualitative and quantitative ways. Experimental results indicate three findings. (1) Larger models experience greater identity drift. (2) Model differences exist, but their effect is not stronger than parameter sizes. (3) Assigning a persona may not help to maintain identity. We hope these three findings can help to improve persona stability in AI-driven dialogue systems, particularly in long-term conversations.
format Preprint
id arxiv_https___arxiv_org_abs_2412_00804
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Examining Identity Drift in Conversations of LLM Agents
Choi, Junhyuk
Hong, Yeseon
Kim, Minju
Kim, Bugeun
Computers and Society
Computation and Language
Large Language Models (LLMs) show impressive conversational abilities but sometimes show identity drift problems, where their interaction patterns or styles change over time. As the problem has not been thoroughly examined yet, this study examines identity consistency across nine LLMs. Specifically, we (1) investigate whether LLMs could maintain consistent patterns (or identity) and (2) analyze the effect of the model family, parameter sizes, and provided persona types. Our experiments involve multi-turn conversations on personal themes, analyzed in qualitative and quantitative ways. Experimental results indicate three findings. (1) Larger models experience greater identity drift. (2) Model differences exist, but their effect is not stronger than parameter sizes. (3) Assigning a persona may not help to maintain identity. We hope these three findings can help to improve persona stability in AI-driven dialogue systems, particularly in long-term conversations.
title Examining Identity Drift in Conversations of LLM Agents
topic Computers and Society
Computation and Language
url https://arxiv.org/abs/2412.00804