Talking to a Know-It-All GPT or a Second-Guesser Claude? How Repair reveals unreliable Multi-Turn Behavior in LLMs

TL;DR

This paper investigates how large language models handle repair in multi-turn dialogues, revealing significant variability and unreliability across different models in responding to user-initiated corrections.

Contribution

It provides the first systematic analysis of repair behavior in LLMs during multi-turn conversations, highlighting model-specific unreliability and susceptibility to manipulation.

Findings

Models differ greatly in their repair responses.

Behavior becomes more unpredictable beyond a single turn.

Each LLM shows unique unreliability patterns.

Abstract

Repair, an important resource for resolving trouble in human-human conversation, remains underexplored in human-LLM interaction. In this study, we investigate how LLMs engage in the interactive process of repair in multi-turn dialogues around solvable and unsolvable math questions. We examine whether models initiate repair themselves and how they respond to user-initiated repair. Our results show strong differences across models: reactions range from being almost completely resistant to (appropriate) repair attempts to being highly susceptible and easily manipulated. We further demonstrate that once conversations extend beyond a single turn, model behavior becomes more distinctive and less predictable across systems. Overall, our findings indicate that each tested LLM exhibits its own characteristic form of unreliability in the context of repair.