[论文解读] AI vs. Human -- Differentiation Analysis of Scientific Content Generation

Yongqiang Ma, Jiawei Liu+5 more|arXiv (Cornell University)|Jan 24, 2023

Topic Modeling被引用 76

一句话总结

本论文构建了一个基于特征的框架，用于区分AI生成与人工撰写的科学摘要，分析连贯性/一致性/论证逻辑，并评估检测方法包括困惑度和一个微调检测器。研究发现写作风格存在差距，AI生成的内容通常外部事实不一致性较少但可能存在事实错误。

ABSTRACT

Recent neural language models have taken a significant step forward in producing remarkably controllable, fluent, and grammatical text. Although studies have found that AI-generated text is not distinguishable from human-written text for crowd-sourcing workers, there still exist errors in AI-generated text which are even subtler and harder to spot. We primarily focus on the scenario in which scientific AI writing assistant is deeply involved. First, we construct a feature description framework to distinguish between AI-generated text and human-written text from syntax, semantics, and pragmatics based on the human evaluation. Then we utilize the features, i.e., writing style, coherence, consistency, and argument logistics, from the proposed framework to analyze two types of content. Finally, we adopt several publicly available methods to investigate the gap of between AI-generated scientific text and human-written scientific text by AI-generated scientific text detection models. The results suggest that while AI has the potential to generate scientific content that is as accurate as human-written content, there is still a gap in terms of depth and overall quality. The AI-generated scientific content is more likely to contain errors in factual issues. We find that there exists a "writing style" gap between AI-generated scientific text and human-written scientific text. Based on the analysis result, we summarize a series of model-agnostic and distribution-agnostic features for detection tasks in other domains. Findings in this paper contribute to guiding the optimization of AI models to produce high-quality content and addressing related ethical and security concerns.

研究动机与目标

• 开发一个特征描述框架，以在语法、语义和语用层面对AI生成与人工撰写的科学文本进行区分。
• 使用写作风格、连贯性、一致性和论证逻辑，在计算机科学和生物领域分析AI与人工科学摘要。
• 评估GPT生成文本的检测方法，包括基于特征的方法和神经模型方法，并具备可解释性。

提出的方法

• 使用GPT-3/Text-Davinci-003生成摘要，提示中包含科学结构信息的优化版本。
• 构建一个四维度的基于特征的检测框架：写作风格、连贯性、一致性和论证逻辑。
• 微调一个GPT-2输出检测器，并与RoBERTa/OpenAI检测器基线进行比较。
• 使用SciBERT基于困惑度进行检测，以设置领域特定阈值（摘要2.6，维基项4.6）。
• 进行人工评估，以评估人类辨别AI生成科学文本的能力并分析相关因素。

实验结果

研究问题

• RQ1 AI生成的科学摘要是否可以通过语法、语义和语用特征来可靠地区分于人工撰写的摘要？
• RQ2 写作风格、连贯性、一致性和论证逻辑对检测性能的贡献如何？
• RQ3 AI生成与人工撰写的科学内容在深度/质量和事实准确性方面存在哪些差距？
• RQ4 困惑度和检测器方法在跨领域识别AI生成的科学文本方面的有效性如何？

主要发现

• 存在显著的写作风格差异（语法）在AI生成与人工撰写的科学文本之间。
• 基于困惑度的检测在摘要上达到较高的F1分数（94%）；在维基项描述上较低（77%）。
• 基于词元级和功能词的特征在语法检测中具有强预测力（逻辑回归解释了高达86.1%的变异）。
• AI生成的摘要与标题高度一致，但内部一致性较低，案例示例中的一些事实引用不正确或被虚构。
• 经过训练的检测模型在区分AI生成与人工撰写的科学文本方面优于人类，支持对科学领域的AI生成内容进行标注。