First several months we were in a full research and experimentation mode: reading papers, testing approaches, combining models and trying to achieve as high accuracy as possible.
Thanks to our deep expertise in deep learning and more than 4 years of experience in NLP by the end of autumn we achieved outstanding results: It’s AI became the most accurate AI detector according to RAID benchmark (the biggest and most robust benchmark in the AI detection field), outperforming detectors like GptZero, Originality, CopyLeaks and others.