What you say or how you say it? Predicting Conflict Outcomes in Real and LLM-Generated Conversations
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916394113695744 |
|---|---|
| author | D'Costa, Priya Ronald Rowbotham, Evan Hu, Xinlan Emily |
| author_facet | D'Costa, Priya Ronald Rowbotham, Evan Hu, Xinlan Emily |
| contents | When conflicts escalate, is it due to what is said or how it is said? In the conflict literature, two theoretical approaches take opposing views: one focuses on the content of the disagreement, while the other focuses on how it is expressed. This paper aims to integrate these two perspectives through a computational analysis of 191 communication features -- 128 related to expression and 63 to content. We analyze 1,200 GPT-4 simulated conversations and 12,630 real-world discussions from Reddit. We find that expression features more reliably predict destructive conflict outcomes across both settings, although the most important features differ. In the Reddit data, conversational dynamics such as turn-taking and conversational equality are highly predictive, but they are not predictive in simulated conversations. These results may suggest a possible limitation in simulating social interactions with language models, and we discuss the implications for our findings on building social computing systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_09338 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | What you say or how you say it? Predicting Conflict Outcomes in Real and LLM-Generated Conversations D'Costa, Priya Ronald Rowbotham, Evan Hu, Xinlan Emily Social and Information Networks Human-Computer Interaction When conflicts escalate, is it due to what is said or how it is said? In the conflict literature, two theoretical approaches take opposing views: one focuses on the content of the disagreement, while the other focuses on how it is expressed. This paper aims to integrate these two perspectives through a computational analysis of 191 communication features -- 128 related to expression and 63 to content. We analyze 1,200 GPT-4 simulated conversations and 12,630 real-world discussions from Reddit. We find that expression features more reliably predict destructive conflict outcomes across both settings, although the most important features differ. In the Reddit data, conversational dynamics such as turn-taking and conversational equality are highly predictive, but they are not predictive in simulated conversations. These results may suggest a possible limitation in simulating social interactions with language models, and we discuss the implications for our findings on building social computing systems. |
| title | What you say or how you say it? Predicting Conflict Outcomes in Real and LLM-Generated Conversations |
| topic | Social and Information Networks Human-Computer Interaction |
| url | https://arxiv.org/abs/2409.09338 |