Google DeepMind 与 UPenn 论文提出可解释的 AI 文本检测器
New Google DeepMind + UPenn paper shows that an AI-text detector can explain its verdicts with readable writing features and still match a black-box detector.
Google DeepMind 与 UPenn 的新论文提出一种可解释的 AI 文本检测器，用可读的写作特征解释判定结果，同时准确率接近黑盒检测器。研究在 Gemma-2-2B 检测器的内部活动上训练稀疏自编码器，把其活动拆分为特征并由 LLM 为每个特征打标签，只保留标签能较好预测其行为的特征输入简单分类器。仅用标注良好的特征，该检测器域内 F1 为 96.8，接近黑盒的 98.0；在未见测试数据上，基于基座模型特征的版本 F1 为 92.2，黑盒为 75.2。
推荐理由
AI 评分 56
论文用稀疏自编码器把黑盒检测器的内部活动拆成可读特征，在保持接近准确率的同时让判定依据可被审查。
主题
正文
New Google DeepMind + UPenn paper shows that an AI-text detector can explain its verdicts with readable writing features and still match a black-box detector.
suggests that anyone deploying AI-text detectors should be able to read and remove the specific cues a detector relies on.
The best AI-text detectors are black-box classifiers, yet they flag student essays and peer reviews. Readable stylometric detectors exist but trail them by wide margins.
They trained sparse autoencoders on a Gemma-2-2B detector's internals to split its activity into features, then had an LLM label each one. They kept only features whose labels predicted their behavior well and fed those into a simple classifier.
With only well-labeled features, the detector kept 96.8 F1 in-domain, close to the black box's 98.0. On unseen test data, a version built on base-model features scored 92.2 F1 against 75.2 for the black box.
If your detector's decisions affect people, build it on labeled features so reviewers can see and drop cues they consider unfair.