analiza aiqbench
This research introduces MemAdapter, a framework designed to mitigate memory-induced sycophancy in LLM agents by leveraging counterfactual reasoning and context-aware reflection. Unlike previous approaches that simply filter biased memories, the method adapts to how valid information can still skew alignment depending on the specific task context. The results demonstrate improved reliability in long-horizon interactions where agents must balance personalization with objective evidence.
Memadapter: Counterfactual Adaptation Against Memory-induced Sycophancy
arXiv:2610.05162v1 Announce Type: new Abstract: Long-term memory enables LLM-based agents to retain and reuse information across tasks and sessions, supporting personalization and long-horizon interactions. However, persistent memories can also induce sycophancy, causing agents to over-align with users' historical beliefs even when they are inaccurate, outdated, or inconsistent with objective evidence. Existing mitigation methods assume that memory-induced sycophancy originates from biased or incorrect memories and attempt to reduce this risk by filtering such memories at different stages of the memory pipeline. However, in the real world, objective and correct memories can still induce sycophancy, and the same memory can warrant different influence across different contexts. To this end, we propose MemAdapter, a novel framework that adaptively integrates retrieved memories to support objective and reliable reasoning. Specifically, MemAdapter consists of three components: (i) Counterfactual Induction, which leverages counterfactual reasoning to uncover the potential risk of retrieved memories; (ii) Context-Aware Reflection, which calibrates the inferential influence of each retrieved memory in light of the current task via self-reflection; and (iii) Evidence-Based Reasoning, which grounds the final response in appropriate evidence while preserving the legitimate influence of memory. Extensive experiments on three benchmarks demonstrate that MemAdapter consistently improves memory reliability across diverse scenarios. Our code is available at https://github.com/DEEP-JLU/MemAdapter.
Abstract:Long-term memory enables LLM-based agents to retain and reuse information across tasks and sessions, supporting personalization and long-horizon interactions. However, persistent memories can also induce sycophancy, causing agents to over-align with users' historical beliefs even when they are inaccurate, outdated, or inconsistent with objective evidence. Existing mitigation methods assume that memory-induced sycophancy originates from biased or incorrect memories and attempt to reduce this risk by filtering such memories at different stages of the memory pipeline. However, in the real world, objective and correct memories can still induce sycophancy, and the same memory can warrant different influence across different contexts. To this end, we propose MemAdapter, a novel framework that adaptively integrates retrieved memories to support objective and reliable reasoning. Specifically, MemAdapter consists of three components: (i) Counterfactual Induction, which leverages counterfactual reasoning to uncover the potential risk of retrieved memories; (ii) Context-Aware Reflection, which calibrates the inferential influence of each retrieved memory in light of the current task via self-reflection; and (iii) Evidence-Based Reasoning, which grounds the final response in appropriate evidence while preserving the legitimate influence of memory. Extensive experiments on three benchmarks demonstrate that MemAdapter consistently improves memory reliability across diverse scenarios. Our code is available at this https URL.
Artificial Intelligence (cs.AI)
arXiv:2610.05162 [cs.AI]
(or arXiv:2610.05162v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.05162 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 296 cuvinte
Subiecte conexe
Value Leakage: An LLM’s Answers Are Silently Shaped by Its Own Values
AI Alignment Forum · 31 iul. 2026
Why do models task game?
AI Alignment Forum · 6 aug. 2026
Proactive Service Agents: A Unified Decision Framework, Methods, and Evaluation
arXiv · 5 sept. 2026
Defense-in-Depth for LLMs: Evaluating Memory Gates Against Activation-Induced and Memory-Induced Sycophancy
arXiv · 7 oct. 2026