{"paper":{"title":"From Human Memory to AI Memory: A Survey on Memory Mechanisms in the Era of LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"This survey connects categories of human memory to memory in LLM-based AI systems and introduces a three-dimension eight-quadrant framework to organize the field.","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Chen Zhang, Huifeng Guo, Ruiming Tang, Sheng Liang, Yaxiong Wu, Yichao Wang, Yong Liu, Yongyue Zhang","submitted_at":"2025-04-22T15:05:04Z","abstract_excerpt":"Memory is the process of encoding, storing, and retrieving information, allowing humans to retain experiences, knowledge, skills, and facts over time, and serving as the foundation for growth and effective interaction with the world. It plays a crucial role in shaping our identity, making decisions, learning from past experiences, building relationships, and adapting to changes. In the era of large language models (LLMs), memory refers to the ability of an AI system to retain, recall, and use information from past interactions to improve future responses and interactions. Although previous res"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"we propose a comprehensive survey on the memory of LLM-driven AI systems. In particular, we first conduct a detailed analysis of the categories of human memory and relate them to the memory of AI systems. Second, we systematically organize existing memory-related work and propose a categorization method based on three dimensions (object, form, and time) and eight quadrants.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That mapping human memory categories directly to AI memory systems and organizing the literature via the proposed three-dimension eight-quadrant scheme will yield actionable insights for building more powerful memory mechanisms in LLMs.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"The paper surveys human memory categories, maps them to LLM memory, and proposes a new three-dimension (object, form, time) categorization into eight quadrants to organize existing work and highlight open problems.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"This survey connects categories of human memory to memory in LLM-based AI systems and introduces a three-dimension eight-quadrant framework to organize the field.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"25e3be365ff61a1501db3bffd21ae6d7d2de9fccdcd985fe3954c17e6520b4d2"},"source":{"id":"2504.15965","kind":"arxiv","version":2},"verdict":{"id":"1392be2b-7bf3-4642-91fc-9aff9faa45da","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T11:00:17.372542Z","strongest_claim":"we propose a comprehensive survey on the memory of LLM-driven AI systems. In particular, we first conduct a detailed analysis of the categories of human memory and relate them to the memory of AI systems. Second, we systematically organize existing memory-related work and propose a categorization method based on three dimensions (object, form, and time) and eight quadrants.","one_line_summary":"The paper surveys human memory categories, maps them to LLM memory, and proposes a new three-dimension (object, form, time) categorization into eight quadrants to organize existing work and highlight open problems.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That mapping human memory categories directly to AI memory systems and organizing the literature via the proposed three-dimension eight-quadrant scheme will yield actionable insights for building more powerful memory mechanisms in LLMs.","pith_extraction_headline":"This survey connects categories of human memory to memory in LLM-based AI systems and introduces a three-dimension eight-quadrant framework to organize the field."},"references":{"count":155,"sample":[{"doi":"","year":2024,"title":"A survey on large language model based autonomous agents","work_id":"b2aac465-d452-4d47-87d9-156aa944bc8c","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Personal LLM Agents: Insights and Survey about the Capability, Efficiency and Security","work_id":"7692e42c-c83d-4527-ba16-9c4b69666d47","ref_index":2,"cited_arxiv_id":"2401.05459","is_internal_anchor":true},{"doi":"","year":2023,"title":"A Survey of Large Language Models","work_id":"de1b42b5-4a0a-4b1f-8c78-1f7fe21be6c9","ref_index":3,"cited_arxiv_id":"2303.18223","is_internal_anchor":true},{"doi":"","year":2024,"title":"A survey on evaluation of large language models","work_id":"abb4c0f1-6bc9-4f60-8dac-91f21de84748","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"All roads lead to rome: Unveiling the trajectory of recommender systems across the llm era","work_id":"c7931a4a-92fd-449d-b717-8df2c0c628e2","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":155,"snapshot_sha256":"3676b7cebf48137207e41164e4a181d609ef71ea202a34b55f8972983bcb1de3","internal_anchors":25},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}