机器生成文本可被识别，因采样策略排除了人类常选的低概率词。
The Truncation Blind Spot: How Decoding Strategies Systematically Exclude Human-Like Token Choices
- 用概率截断策略限制生成，忽略人类常用但低概率的词。
- 8–18%人类选词落在常见截断范围外，实词排除率是虚词的2.9倍。
- 降低截断强度可减少可检测性，适合研究生成质量与可识别性者看。
为何机器生成文本仍易被识别？我们追溯至解码阶段：标准策略如top-k和核采样仅限于高概率词，而人类作者常选择语境恰当但模型概率较低的词。截断导致可观测比例的人类词汇无法被生成，我们称之为‘截断盲区’。在五个开源模型和三个领域中，8–18%的人类选词超出常见截断边界，内容词被排除的概率是虚词的2.9倍。因此，仅用两个特征的简单分类器即可在平均AUC-ROC达0.97的情况下区分180万条机器生成与人类文本，且可检测性随截断强度上升，而非模型规模、架构或对齐程度。概率底限采样显著缩小盲区，生成实验也证实了排除与可检测性的关联。代码、数据及测量工具已公开于https://github.com/EstebanGarces/human_vs_machine。
原文摘要 · Abstract (English)
Why does machine-generated text remain detectable? We trace the answer to the decoding stage: standard strategies such as top-$k$ and nucleus sampling restrict generation to high-probability tokens, while human writers routinely choose words that are contextually apt yet rank low under a model's next-token distribution. Truncation therefore renders a measurable share of human word choices unreachable; we call this set the truncation blind spot. Across five open models and three domains, 8--18\% of human-selected tokens fall outside common truncation boundaries, with content words excluded at 2.9 times the rate of function words. As a consequence, simple classifiers using two features separate 1.8 million machine generations from human text at mean AUC-ROC near 0.97, and detectability tracks truncation intensity rather than model scale, architecture, or alignment. Probability-floor samplers substantially narrow the blind spot, and generation experiments confirm the exclusion--detectability link out of sample. Code, data, and the measurement harness are released at https://github.com/EstebanGarces/human_vs_machine.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。