ChatGPT language recognition limitations Prompts in Hungarian in a network of Latin and Cyrillic letters (Netlinguistics case study. The “Chinese room” experiment)

Abstract

The net linguistics case study points to the characteristic character set-focused language recognition limitations of the ChatGPT generative artificial intelligence using prompts in Hungarian but coded by Cyrillic letters. The colloquial English language of its central text enables the objective presentation of message exchanges with AI, showing hybrid characteristics and drawing scientific conclusions. The research is based on a written communication experiment, and its results are presented and interpreted with the help of artificial intelligence. Thus, while AI’s language identification and corresponding message processing limitations are the research’s subject, they also contribute to illuminating the causes and context of the phenomenon. The findings in the study cannot be generalised to other large language models, artificial intelligence, or other linguistic or coding relations different from those investigated.