MemTrapBench: Benchmarking Cognitive Traps in LLM Memory Use
[Sztuczna inteligencja] Autorzy: Mengru Wang, Haozhe Luo, Zhenqian Xu i 6 więcej
Memory has become a key component of large language models, enabling them to retain information and learn from long-term interactions. However, existing memory benchmarks mainly evaluate whether information is correctly extracted, stored, and retrieved, while largely overlooking how retrieved memories reshape model reasoning and affect performance on the current task. We identify memory-induced cognitive traps: even faithfully recorded and semantically relevant memories can distort model reasoning or beliefs and degrade current task performance. To systematically evaluate these failure modes, we introduce MemTrapBench, which covers two forms of cognitive traps: Reasoning Fixation and Belief Distortion. Experiments across two model families and five representative memory frameworks show that MemTrapBench is challenging: all evaluated memory strategies underperform the no-memory setting, …
- Kategoria arxiv
- cs.AI
- Kategoria
- Sztuczna inteligencja