Robust and Fine-Grained Detection of AI Generated Texts
This addresses the need for robust detection systems as advanced LLMs proliferate, though it is incremental by focusing on fine-grained and co-authored cases.
The paper tackles the problem of detecting AI-generated content, especially in short and human-LLM co-authored texts, by introducing token classification models trained on a dataset of over 2.4M texts across 23 languages, which performed well on unseen domains, generators, and adversarial inputs.
An ideal detection system for machine generated content is supposed to work well on any generator as many more advanced LLMs come into existence day by day. Existing systems often struggle with accurately identifying AI-generated content over shorter texts. Further, not all texts might be entirely authored by a human or LLM, hence we focused more over partial cases i.e human-LLM co-authored texts. Our paper introduces a set of models built for the task of token classification which are trained on an extensive collection of human-machine co-authored texts, which performed well over texts of unseen domains, unseen generators, texts by non-native speakers and those with adversarial inputs. We also introduce a new dataset of over 2.4M such texts mostly co-authored by several popular proprietary LLMs over 23 languages. We also present findings of our models' performance over each texts of each domain and generator. Additional findings include comparison of performance against each adversarial method, length of input texts and characteristics of generated texts compared to the original human authored texts.