The heads of the International Machine Tuition Conference (ICML) prohibited the articles created using large language models ( LLM), such as Chatgpt.
Documents containing the generated text are prohibited if the created text is not part of the experimental analysis of the article. However, scientists are allowed to use AI to adjust their text to improve its style or grammar. ICML academics in order to protect against plagiarism banned the reception of articles created using AI. It is worth noting that these rules are not legally fixed, so they can still change in the future.
Whether scientists decide to adhere to the rules or not, depends on them. At the moment, there are no tools that effectively detect the generated text, so ICML will rely only on people who note suspicious documents during the verification process. The generated text often contains actual errors, and the authors will have to edit the text of AI so as not to cause suspicion.
Language models, such as ChatGPT, study on the text from the Internet. They learn to find general patterns between words in order to predict what to write further, given the text instruction or hint. The question is whether these systems are engaged in plagiarism or not? Now there is no evidence that ChatGPT directly copies the text of someone else’s article, but its results are based on a letter from people. If he simply copies the text, is the machine letter the subject of copyright?
Earlier, the Department of Education of New York banned banned the use of the ChatGPT chat boot on devices belonging to the Department of Education.