TAME: A Trustworthy Test-Time Evolution of Agent Memory with Systematic Benchmarking
🤖 AI Summary
This work addresses the problem of “memory mis-evolution” in intelligent agents during test-time memory adaptation, which can degrade trustworthiness and compromise safety and alignment—even under benign task evolution. To mitigate this issue, the paper introduces TAME, a novel dual-track memory evolution framework that separately optimizes executor and evaluator memories. TAME employs a closed-loop process comprising memory filtering, draft generation, trustworthy refinement, execution feedback, and dual-track updating to enable coordinated memory evolution. Evaluated on the newly proposed Trust-Memevo benchmark—a systematic suite for assessing agent trustworthiness—TAME effectively alleviates memory mis-evolution across diverse task domains, simultaneously enhancing task performance and significantly improving agent trustworthiness and safety.
Application Category
📝 Abstract
Test-time evolution of agent memory serves as a pivotal paradigm for achieving AGI by bolstering complex reasoning through experience accumulation. However, even during benign task evolution, agent safety alignment remains vulnerable-a phenomenon known as Agent Memory Misevolution. To evaluate this phenomenon, we construct the Trust-Memevo benchmark to assess multi-dimensional trustworthiness during benign task evolution, revealing an overall decline in trustworthiness across various task domains and evaluation settings. To address this issue, we propose TAME, a dual-memory evolutionary framework that separately evolves executor memory to improve task performance by distilling generalizable methodologies, and evaluator memory to refine assessments of both safety and task utility based on historical feedback. Through a closed loop of memory filtering, draft generation, trustworthy refinement, execution, and dual-track memory updating, TAME preserves trustworthiness without sacrificing utility. Experiments demonstrate that TAME mitigates misevolution, achieving a joint improvement in both trustworthiness and task performance.
Problem
Research questions and friction points this paper is trying to address.
Agent Memory Misevolution
Test-time Evolution
Trustworthiness
Safety Alignment
AGI
Innovation
Methods, ideas, or system contributions that make the work stand out.
Test-time evolution
Dual-memory framework
Agent Memory Misevolution
Trustworthiness
AGI safety
🔎 Similar Papers
No similar papers found.
