Amnesia by Design, Memory By Necessity: Persistent State for Document Intelligence
Not on the map. Why?
Abstract (arXiv)
Modern Document AI reads contracts, extracts fields, reasons over tables, and grounds answers to page regions, then forgets everything. Processing an amendment the next day begins from scratch: no schema retained, no contradiction detected, no experience carried forward. This is a structural choice, not a scale failure: current systems are stateless functions. We call this the statelessness bottleneck. This bottleneck lies beyond parameter scaling, context extension, and retrieval augmentation: storage provides persistence and retrieval provides access, but neither consolidates observations into knowledge that improves future processing. This survey formalizes persistent evidence-grounded document state as a unifying framework, specifying the operations and invariants required to convert multimodal evidence into durable, provenance-linked state. We introduce a statefulness audit showing that ten representative benchmarks, coded against eight statefulness criteria, leave cross-session state evolution untested, and derive a longitudinal benchmark harness with five counterfactual metrics: Experience Gain, Cost Efficiency, Memory Harm, Forgetting Fidelity, Coverage Retention, to characterize the benefit, cost, risk, and governability of persistent document state. Document AI lacks mechanisms coupling persistent state to document-native structure, provenance, and temporal validity. The next era of Document AI will be defined by what systems retain across documents, sessions, and time.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 140 of its 183 references that are in the map. First posted on arXiv 25 Sep 2026.
Builds on
140 of 183 references.
References · 140
- 2017Attention Is All You Needcited by 59,932
- 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
- 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,160
- 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
- 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,767
- 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
- 2023Mamba: Linear-Time Sequence Modeling with Selective State Spacescited by 4,616
- 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,581
- 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,072
- 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,064
- 2020Dense Passage Retrieval for Open-Domain Question Answeringcited by 3,779
- 2022FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awarenesscited by 3,611
38 more
- 2023Generative Agents: Interactive Simulacra of Human Behaviorcited by 3,500
- 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,383
- 2020Longformer: The Long-Document Transformercited by 3,362
- 2016iCaRL: Incremental Classifier and Representation Learningcited by 2,313
- 2023Retrieval-Augmented Generation for Large Language Models: A Surveycited by 2,185
- 2020REALM: Retrieval-Augmented Language Model Pre-Trainingcited by 2,126
- 2023Efficient Streaming Language Models with Attention Sinkscited by 2,115
- 2023Voyager: An Open-Ended Embodied Agent with Large Language Modelscited by 2,081
- 2017Gradient Episodic Memory for Continual Learningcited by 1,871
- 2019Transformer-XL: Attentive Language Models Beyond a Fixed-Length Contextcited by 1,830
- 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,800
- 2019Towards VQA Models That Can Readcited by 1,696
- 2020Big Bird: Transformers for Longer Sequencescited by 1,673
- 2020Reformer: The Efficient Transformercited by 1,658
- 2019Generating Long Sequences with Sparse Transformerscited by 1,523
- 2022ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoningcited by 1,449
- 2024From Local to Global: A Graph RAG Approach to Query-Focused Summarizationcited by 1,411
- 2018Continual Lifelong Learning with Neural Networks: A Reviewcited by 1,313
- 2023MemGPT: Towards LLMs as Operating Systemscited by 1,295
- 2020DocVQA: A Dataset for VQA on Document Imagescited by 1,288
- 2021Improving language models by retrieving from trillions of tokenscited by 1,278
- 2020ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERTcited by 1,160
- 2021Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolationcited by 1,085
- 2019Machine Unlearningcited by 1,020
- 2022Atlas: Few-shot Learning with Retrieval Augmented Language Modelscited by 940
- 2023Large Language Models Can Be Easily Distracted by Irrelevant Contextcited by 820
- 2024Evaluating Very Long-Term Conversational Memory of LLM Agentscited by 800
- 2025Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memorycited by 722
- 2023Extending Context Window of Large Language Models via Positional Interpolationcited by 721
- 2024LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memorycited by 656
- 2024A Survey on the Memory Mechanism of Large Language Model based Agentscited by 622
- 2023YaRN: Efficient Context Window Extension of Large Language Modelscited by 584
- 2021FinQA: A Dataset of Numerical Reasoning over Financial Datacited by 552
- 2019LayoutLM: Pre-training of Text and Layout for Document Image Understandingcited by 549
- 2023MemoryBank: Enhancing Large Language Models with Long-Term Memorycited by 539
- 2015An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognitioncited by 519
- 2022LayoutLMv3: Pre-training for Document AI with Unified Text and Image Maskingcited by 481
- 2021A Dataset of Information-Seeking Questions and Answers Anchored in Research Paperscited by 475
50 most cited of 140 All 140 (JSON)
Built on by
Not cited yet.