# Drift No More? Context Equilibria in Multi-Turn LLM Interactions

arXiv: https://arxiv.org/abs/2510.07777 · HTML: https://mlsanity.com/paper/2510.07777/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 37 references that are in the map.
- First posted on arXiv 9 Oct 2025.

## Builds on (most cited first, 23 of 23)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Survey of Hallucination in Natural Language Generation](https://mlsanity.com/paper/2202.03629/) (2022)
- [τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains](https://mlsanity.com/paper/2406.12045/) (2024)
- [LaMDA: Language Models for Dialog Applications](https://mlsanity.com/paper/2201.08239/) (2022)
- [LIMA: Less Is More for Alignment](https://mlsanity.com/paper/2305.11206/) (2023)
- [LongLLMLingua: Accelerating and Enhancing LLMs in Long Context Scenarios via Prompt Compression](https://mlsanity.com/paper/2310.06839/) (2023)
- [LLMs Get Lost In Multi-Turn Conversation](https://mlsanity.com/paper/2505.06120/) (2025)
- [Describe, Explain, Plan and Select: Interactive Planning with Large Language Models Enables Open-World Multi-Task Agents](https://mlsanity.com/paper/2302.01560/) (2023)
- [AgentBoard: An Analytical Evaluation Board of Multi-turn LLM Agents](https://mlsanity.com/paper/2401.13178/) (2024)
- [MINT: Evaluating LLMs in Multi-turn Interaction with Tools and Language Feedback](https://mlsanity.com/paper/2309.10691/) (2023)
- [Long-context LLMs Struggle with Long In-context Learning](https://mlsanity.com/paper/2404.02060/) (2024)
- [Deal or No Deal? End-to-End Learning for Negotiation Dialogues](https://mlsanity.com/paper/1706.05125/) (2017)
- [How Well Can LLMs Negotiate? NegotiationArena Platform and Analysis](https://mlsanity.com/paper/2402.05863/) (2024)
- [Evaluating LLM-based Agents for Multi-Turn Conversations: A Survey](https://mlsanity.com/paper/2503.22458/) (2025)
- [Measuring and Controlling Instruction (In)Stability in Language Model Dialogs](https://mlsanity.com/paper/2402.10962/) (2024)
- [What's the Magic Word? A Control Theory of LLM Prompting](https://mlsanity.com/paper/2310.04444/) (2023)
- [Interactive Task Planning with Language Models](https://mlsanity.com/paper/2310.10645/) (2023)
- [Goal Alignment in LLM-Based User Simulators for Conversational AI](https://mlsanity.com/paper/2507.20152/) (2025)
- [ReSpAct: Harmonizing Reasoning, Speaking, and Acting Towards Building Large Language Model-Based Conversational AI Agents](https://mlsanity.com/paper/2411.00927/) (2024)
- [MIRAGE: A Benchmark for Multimodal Information-Seeking and Reasoning in Agricultural Expert-Guided Conversations](https://mlsanity.com/paper/2506.20100/) (2025)

## Built on by (most cited first, 22 of 22)

- [BeliefShift: Benchmarking Temporal Belief Consistency and Opinion Drift in LLM Agents](https://mlsanity.com/paper/2603.23848/) (2026)
- [Inherited Goal Drift: Contextual Pressure Can Undermine Agentic Goals](https://mlsanity.com/paper/2603.03258/) (2026)
- [Graph of States: Solving Abductive Tasks with Large Language Models](https://mlsanity.com/paper/2603.21250/) (2026)
- [ContextEcho: A Benchmark for Persona Drift in Long Agentic-Coding Sessions](https://mlsanity.com/paper/2605.24279/) (2026)
- [The Narrative Continuity Test: A Conceptual Framework for Evaluating Identity Persistence in AI Systems](https://mlsanity.com/paper/2510.24831/) (2025)
- [AI Planning Framework for LLM-Based Web Agents](https://mlsanity.com/paper/2603.12710/) (2026)
- [Models Recall What They Violate: Constraint Adherence in Multi-Turn LLM Ideation](https://mlsanity.com/paper/2604.28031/) (2026)
- [Diagnosing and Mitigating Context Rot in Long-horizon Search](https://mlsanity.com/paper/2606.29718/) (2026)
- [Multi2: Hierarchical Multi-Agent Decision-Making with LLM-Based Agents in Interactive Environments](https://mlsanity.com/paper/2606.03698/) (2026)
- [Hallucination as Context Drift: Synchronization Protocols for Multi-Agent LLM Systems](https://mlsanity.com/paper/2606.21666/) (2026)
- [Enforcing Monotonic Progress in Legal Cross-Examination: Preventing Long-Horizon Stagnation in LLM-Based Inquiry](https://mlsanity.com/paper/2602.04206/) (2026)
- [AMEL: Accumulated Message Effects on LLM Judgments](https://mlsanity.com/paper/2605.22714/) (2026)
- [Controllable Memory Usage: Balancing Anchoring and Innovation in Long-Term Human-Agent Interaction](https://mlsanity.com/paper/2601.05107/) (2026)
- [Toward Measuring Structural Drift in LLM Communication Loops](https://mlsanity.com/paper/2604.13061/) (2026)
- [Coordinates of Capability: A Unified MTMM-Geometric Framework for LLM Evaluation](https://mlsanity.com/paper/2605.08522/) (2026)
- [From Context Shift to Stylistic Collapse: Why Training Objectives Matter More Than Scale](https://mlsanity.com/paper/2605.28826/) (2026)
- [Measuring What Persists: Conditioning Mechanisms and a Geometric Framework for AI Agent Identity](https://mlsanity.com/paper/2606.21843/) (2026)
- [Semantic Drift and the Stability of Operator Control in Reasoning-Class Decision Support Systems](https://mlsanity.com/paper/2607.09790/) (2026)
- [Relational Positioning as a Measurable Risk Object: History-Carried Lock-in and Self-Confabulation in Multi-Turn Human-AI Dialogue](https://mlsanity.com/paper/2607.11437/) (2026)
- [Evaluating Large Language Models on Misconceptions in Multi-Turn Medical Conversations](https://mlsanity.com/paper/2607.12884/) (2026)
- [SpecFirst: Behavioral Specification Elicitation as a First-Class Step in Agent-Based Program Synthesis from Scratch](https://mlsanity.com/paper/2607.27167/) (2026)
- [Model vs. User: Whose Values Prevail? Inference-Time Value Alignment in Multi-Turn Conversations](https://mlsanity.com/paper/2610.19817/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) excel at single-turn tasks such as instruction following and summarization, yet real-world deployments require sustained multi-turn interactions where user goals and conversational context persist and evolve. A recurring challenge in this setting is context drift: the gradual divergence of a model's outputs from goal-consistent behavior across turns. Unlike single-turn errors, drift unfolds temporally and is poorly captured by static evaluation metrics. In this work, we present a study of context drift in multi-turn interactions and propose a simple dynamical framework to interpret its behavior. We formalize drift as the turn-wise KL divergence between the token-level predictive distributions of the test model and a goal-consistent reference model, and propose a recurrence model that interprets its evolution as a bounded stochastic process with restoring forces and controllable interventions. We instantiate this framework in both synthetic long-horizon rewriting tasks and realistic user-agent simulations such as in τ-Bench, measuring drift for several open-weight LLMs that are used as user simulators. Our experiments consistently reveal stable, noise-limited equilibria rather than runaway degradation, and demonstrate that simple reminder interventions reliably reduce divergence in line with theoretical predictions. Together, these results suggest that multi-turn drift can be understood as a controllable equilibrium phenomenon rather than as inevitable decay, providing a foundation for studying and mitigating context drift in extended interactions.