How well can machine-generated texts be identified and can language models be trained to avoid identification?
This addresses the challenge of distinguishing human from machine text for applications like content moderation, with incremental improvements in evasion techniques.
The study tackled the problem of identifying machine-generated texts, finding that shallow learning classifiers achieve 0.6-0.8 accuracy, while transformer-based classifiers reach over 0.9 accuracy, and a reinforcement learning approach reduced detection to 0.15 or less.
With the rise of generative pre-trained transformer models such as GPT-3, GPT-NeoX, or OPT, distinguishing human-generated texts from machine-generated ones has become important. We refined five separate language models to generate synthetic tweets, uncovering that shallow learning classification algorithms, like Naive Bayes, achieve detection accuracy between 0.6 and 0.8. Shallow learning classifiers differ from human-based detection, especially when using higher temperature values during text generation, resulting in a lower detection rate. Humans prioritize linguistic acceptability, which tends to be higher at lower temperature values. In contrast, transformer-based classifiers have an accuracy of 0.9 and above. We found that using a reinforcement learning approach to refine our generative models can successfully evade BERT-based classifiers with a detection accuracy of 0.15 or less.