English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

With the advent of fluent generative language models that can produceconvincing utterances very similar to those written by humans, distinguishingwhether a piece of text is machine-generated or human-written becomes morechallenging and more important, as such models could be used to spreadmisinformation, fake news, fake reviews and to mimic certain authors andfigures. To this end, there have been a slew of methods proposed to detectmachine-generated text. Most of these methods need access to the logits of thetarget model or need the ability to sample from the target. One such black-boxdetection method relies on the observation that generated text is locallyoptimal under the likelihood function of the generator, while human-writtentext is not. We find that overall, smaller and partially-trained models arebetter universal text detectors: they can more precisely detect text generatedfrom both small and larger models. Interestingly, we find that whether thedetector and generator were trained on the same data is not criticallyimportant to the detection success. For instance the OPT-125M model has an AUCof 0.81 in detecting ChatGPT generations, whereas a larger model from the GPTfamily, GPTJ-6B, has AUC of 0.45.

Smaller Language Models are Better Black-box Machine-Generated Text Detectors