Personalized memory updates improve long term ai assistant recall
Personalized Memory Rubrics: Learning What to Remember per User via Rubric-Guided Credit Assignment
Artificial Intelligence
Summary
Long-term AI assistants need to remember what is important for each user, but deciding what to keep or change is difficult. The authors developed a system called Personalized Memory Rubrics (PMR) that helps AI learn better memory updates by giving detailed feedback tailored to each user. This feedback breaks down overall success into small steps and adjusts to individual users. The method improves how AI remembers information without slowing it down when it is used.
What this means in practice
- •For virtual assistant developers: Improve long-term memory management in AI assistants by tailoring update decisions to individual users for better personalized interactions.
- •For recommendation system engineers: Enhance personalized recommendation algorithms by applying fine-grained, user-specific feedback to memory and decision processes over time.
Authors
Yupeng Han, Kai Zhang, Shuochen Liu, Ze Liu, Xianquan Wang, Zhihong Pan, Xiangyu Zhao, Qi Liu, Enhong Chen
Abstract
Long-horizon personalized agents maintain an evolving memory by continuously deciding what to add, modify, or discard. However, learning an optimal update policy remains challenging. Conventional training relies solely on terminal question-answering accuracy, yielding a coarse feedback signal that treats all intermediate updates identically. Furthermore, update utility is inherently user-specific, as identical memory actions may yield opposite effects across different user personas. Existing methods fail to address this heterogeneity, relying on globally shared evaluation standards with at most a single scalar intermediate reward. To address these challenges, we propose \textbf{Personalized Memory Rubrics} (\textbf{PMR}), a framework that translates terminal feedback into fine-grained, per-step, and per-user credit assignment. PMR decomposes evaluation into a universal layer capturing cross-persona failure patterns and a per-user adaptive layer modeling individual traits as a specialized delta over the universal baseline. Through \textbf{Rubric-Guided Credit Assignment}, PMR redistributes credit at the token level, selectively amplifying validated decisions on successful trajectories while falling back to standard policy optimization on failed ones to avoid unreliable attribution. The rubric and memory policy \textbf{co-evolve} in a closed loop with anti-circularity guarantees. Crucially, PMR incurs zero inference overhead, operating without any rubric or external judge during deployment. Extensive experiments across four benchmarks demonstrate consistent gains over strong memory-augmented baselines.