Mihály Kiss, a master’s student in software engineering at the University of Szeged (SZTE), has developed an innovative AI-powered text recognition tool. This tool can accurately determine whether text is human-written or generated by large language models, including Hungarian texts. Established in mid-2023 amidst the rise of AI writing tools like ChatGPT, the project emerged as a critical response to concerns about AI’s impact on education and beyond.
Kiss’s detector analyzes patterns unique to large language models, overcoming challenges posed by a lack of suitable Hungarian-language training data. He compiled a diverse dataset of over 350,000 texts from various sources, ensuring comprehensive language exposure. The AI model achieved an impressive accuracy of 0.98 in comparative tests, far exceeding competitors. Publicly accessible, the tool can process short texts almost instantly and holds potential applications in education, media, and legal contexts. Kiss’s work earned him the SZTE Student Innovation Award.
Source link
