On the Zero-Shot Generalization of Machine-Generated Text Detectors

TL;DR

This paper investigates how machine-generated text detectors perform on unseen language models, finding that detectors trained on medium-sized models can effectively generalize to larger models in a zero-shot manner.

Contribution

It introduces a comprehensive evaluation of detector generalization across diverse language models and demonstrates the effectiveness of ensemble training on medium-sized models for robust detection.

Findings

Detectors trained on medium-sized models can zero-shot generalize to larger models.

Ensemble training on medium-sized models enhances detector robustness.

Detectors do not generalize well across all generators, highlighting the challenge of unseen model detection.

Abstract

The rampant proliferation of large language models, fluent enough to generate text indistinguishable from human-written language, gives unprecedented importance to the detection of machine-generated text. This work is motivated by an important research question: How will the detectors of machine-generated text perform on outputs of a new generator, that the detectors were not trained on? We begin by collecting generation data from a wide range of LLMs, and train neural detectors on data from each generator and test its performance on held-out generators. While none of the detectors can generalize to all generators, we observe a consistent and interesting pattern that the detectors trained on data from a medium-size LLM can zero-shot generalize to the larger version. As a concrete application, we demonstrate that robust detectors can be built on an ensemble of training data from medium-sized models.