Feature-Level Insights into Artificial Text Detection with Sparse Autoencoders

• URL: http://arxiv.org/abs/2503.03601v1
• Date: Wed, 05 Mar 2025 15:33:52 GMT
• Title: Feature-Level Insights into Artificial Text Detection with Sparse Autoencoders
• Authors: Kristian Kuznetsov, Laida Kushnareva, Polina Druzhinina, Anton Razzhigaev, Anastasia Voznyuk, Irina Piontkovskaya, Evgeny Burnaev, Serguei Barannikov,
• Abstract summary: We use Sparse Autoencoders (SAE) to extract features from Gemma-2-2b residual stream.<n>We identify both interpretable and efficient features, analyzing their semantics and relevance.<n>Our methods offer valuable insights into how texts from various models differ from human-written content.
• Score: 20.557610461777344
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Artificial Text Detection (ATD) is becoming increasingly important with the rise of advanced Large Language Models (LLMs). Despite numerous efforts, no single algorithm performs consistently well across different types of unseen text or guarantees effective generalization to new LLMs. Interpretability plays a crucial role in achieving this goal. In this study, we enhance ATD interpretability by using Sparse Autoencoders (SAE) to extract features from Gemma-2-2b residual stream. We identify both interpretable and efficient features, analyzing their semantics and relevance through domain- and model-specific statistics, a steering approach, and manual or LLM-based interpretation. Our methods offer valuable insights into how texts from various models differ from human-written content. We show that modern LLMs have a distinct writing style, especially in information-dense domains, even though they can produce human-like outputs with personalized prompts.

Related papers

• Human Variability vs. Machine Consistency: A Linguistic Analysis of Texts Generated by Humans and Large Language Models [0.0]
  We identify significant differences between human-written texts and those generated by large language models (LLMs)<n>Our findings indicate that humans write texts that are less cognitively demanding, with higher semantic content, and richer emotional content compared to texts generated by LLMs.
  arXiv  Detail & Related papers  (2024-12-04T04:38:35Z)
• GigaCheck: Detecting LLM-generated Content [72.27323884094953]
  In this work, we investigate the task of generated text detection by proposing the GigaCheck. Our research explores two approaches: (i) distinguishing human-written texts from LLM-generated ones, and (ii) detecting LLM-generated intervals in Human-Machine collaborative texts. Specifically, we use a fine-tuned general-purpose LLM in conjunction with a DETR-like detection model, adapted from computer vision, to localize AI-generated intervals within text.
  arXiv  Detail & Related papers  (2024-10-31T08:30:55Z)
• Unveiling Large Language Models Generated Texts: A Multi-Level Fine-Grained Detection Framework [9.976099891796784]
  Large language models (LLMs) have transformed human writing by enhancing grammar correction, content expansion, and stylistic refinement. Existing detection methods, which mainly rely on single-feature analysis and binary classification, often fail to effectively identify LLM-generated text in academic contexts. We propose a novel Multi-level Fine-grained Detection framework that detects LLM-generated text by integrating low-level structural, high-level semantic, and deep-level linguistic features.
  arXiv  Detail & Related papers  (2024-10-18T07:25:00Z)
• Detecting Machine-Generated Long-Form Content with Latent-Space Variables [54.07946647012579]
  Existing zero-shot detectors primarily focus on token-level distributions, which are vulnerable to real-world domain shifts. We propose a more robust method that incorporates abstract elements, such as event transitions, as key deciding factors to detect machine versus human texts.
  arXiv  Detail & Related papers  (2024-10-04T18:42:09Z)
• Towards Reliable Detection of LLM-Generated Texts: A Comprehensive Evaluation Framework with CUDRT [9.682499180341273]
  Large language models (LLMs) have significantly advanced text generation, but the human-like quality of their outputs presents major challenges.<n>We propose CUDRT, a comprehensive evaluation framework and bilingual benchmark in Chinese and English.<n>This framework supports scalable, reproducible experiments and enables analysis of how operational diversity, multilingual training sets, and LLM architectures influence detection performance.
  arXiv  Detail & Related papers  (2024-06-13T12:43:40Z)
• Spotting AI's Touch: Identifying LLM-Paraphrased Spans in Text [61.22649031769564]
  We propose a novel framework, paraphrased text span detection (PTD) PTD aims to identify paraphrased text spans within a text. We construct a dedicated dataset, PASTED, for paraphrased text span detection.
  arXiv  Detail & Related papers  (2024-05-21T11:22:27Z)
• Towards Next-Generation Steganalysis: LLMs Unleash the Power of Detecting Steganography [18.7168443402118]
  Linguistic steganography provides convenient implementation to hide messages, particularly with the emergence of AI generation technology. Existing methods are limited to finding distribution differences between steganographic texts and normal texts from the aspect of symbolic statistics. This paper propose to employ human-like text processing abilities of large language models (LLMs) to realize the difference from the aspect of human perception.
  arXiv  Detail & Related papers  (2024-05-15T04:52:09Z)
• A Simple yet Efficient Ensemble Approach for AI-generated Text Detection [0.5840089113969194]
  Large Language Models (LLMs) have demonstrated remarkable capabilities in generating text that closely resembles human writing. It is essential to build automated approaches capable of distinguishing between artificially generated text and human-authored text. We propose a simple yet efficient solution by ensembling predictions from multiple constituent LLMs.
  arXiv  Detail & Related papers  (2023-11-06T13:11:02Z)
• Successor Features for Efficient Multisubject Controlled Text Generation [48.37713738712319]
  We introduce SF-GEN, which is grounded in two primary concepts: successor features (SFs) and language model rectification. SF-GEN seamlessly integrates the two to enable dynamic steering of text generation with no need to alter the LLM's parameters. To the best of our knowledge, our research represents the first application of successor features in text generation.
  arXiv  Detail & Related papers  (2023-11-03T00:17:08Z)
• Towards LLM-guided Causal Explainability for Black-box Text Classifiers [16.36602400590088]
  We aim to leverage the instruction-following and textual understanding capabilities of recent Large Language Models to facilitate causal explainability. We propose a three-step pipeline via which, we use an off-the-shelf LLM to identify the latent or unobserved features in the input text. We experiment with our pipeline on multiple NLP text classification datasets, and present interesting and promising findings.
  arXiv  Detail & Related papers  (2023-09-23T11:22:28Z)
• Towards Codable Watermarking for Injecting Multi-bits Information to LLMs [86.86436777626959]
  Large language models (LLMs) generate texts with increasing fluency and realism. Existing watermarking methods are encoding-inefficient and cannot flexibly meet the diverse information encoding needs. We propose Codable Text Watermarking for LLMs (CTWL) that allows text watermarks to carry multi-bit customizable information.
  arXiv  Detail & Related papers  (2023-07-29T14:11:15Z)
• Harnessing Explanations: LLM-to-LM Interpreter for Enhanced Text-Attributed Graph Representation Learning [51.90524745663737]
  A key innovation is our use of explanations as features, which can be used to boost GNN performance on downstream tasks. Our method achieves state-of-the-art results on well-established TAG datasets. Our method significantly speeds up training, achieving a 2.88 times improvement over the closest baseline on ogbn-arxiv.
  arXiv  Detail & Related papers  (2023-05-31T03:18:03Z)
• MAGE: Machine-generated Text Detection in the Wild [82.70561073277801]
  Large language models (LLMs) have achieved human-level text generation, emphasizing the need for effective AI-generated text detection. We build a comprehensive testbed by gathering texts from diverse human writings and texts generated by different LLMs. Despite challenges, the top-performing detector can identify 86.54% out-of-domain texts generated by a new LLM, indicating the feasibility for application scenarios.
  arXiv  Detail & Related papers  (2023-05-22T17:13:29Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.