A curious publication appeared in the New York Times: American students began to cheat on artificial intelligence. Several professors from different universities began to notice that student papers – essays, reports, studies, answers to exam questions – are written extremely smoothly, impeccably reasoned and convincingly. On the one hand, there is only to rejoice, on the other hand, doubts were caused by the seemingly equal level of horizons in the selection of arguments (and extremely broad, almost at the level of an academician) and the structure as if according to a textbook.
We conducted a small study-investigation - and found out that such essays were written off from the answers to questions that the ChatGPT chatbot gave. Its development is already so good that for the questions asked it is able to give a very coherent, meaningful, "smooth" text, practically indistinguishable from real human speech.
ChatGPT is a development of the American company OpenAI: as the name implies, its specialization is just working with artificial intelligence. OpenAI has already managed to create DALL-E 2 (creating images from description) and GPT-3 (bot for programming).
ChatGPT has already been blocked in American educational institutions (at least in internal computer networks), and OpenAI said it would develop a "response to Chamberlain" – a program that will determine whether a person or a bot created the text.
In general, the situation is quite innovative, so the methods of solving the problem will have to invent new ones. In particular, it will be necessary to start with ethical codes: it has always been forbidden to write off a person, but the concept of "robot" or "AI" was simply not included in such prohibitions, there was no sense and technical possibility. Teachers are beginning to change their working methods: for example, they ask fewer written essays, and more focus on classroom classes, oral answers, group projects. They try to use for work and analysis texts fresher, newer, and not conventional Shakespeare, with which the Internet and bots around the world are already familiar from start to finish. Some suggest revising the evaluation criteria, making them stricter, but this is a very heterogeneous and multifaceted issue, so the scale of assessments is unlikely to change in the near future.
It is noteworthy that some teachers and professors see in bots and artificial intelligence new opportunities for learning! For example, they can generate answers, and then discuss in class: how much the "opinion" of the robot differs / converges with the opinion of a person. You can look for inaccuracies and errors in "artificial" answers (in particular, AI often misrepresents quotes and sources of information): finding an error, the student will show that he himself knows the correct answer.
