In Europe and America universities the first cases of writing student papers have already been identified using ChatGPT software. In Germany, the historic University of Tübingen has decided to limit the use of this software to students and researchers. In the US, professors verbally check students on the content of the assignments to make sure they know the details and whether they were prepared by ChatGPT. New York Public Schools banned its use altogether.

ChatGPT, an artificial intelligence application launched last December, in a very short time, he showed that he would introduce new demons into university teaching, research work and examinations, admit university officials to “K”. The speed of distribution and adoption of chatbots is amazing.

“Can the logic of a universal ban in this new reality solve the problem? Can we talk about the banning of such applications in the academic environment that conducts research aimed at the further development of these applications? Do not believe. Amending the rules of ethics and ethics enough? There are many aspects of the question, and the answers are not at all obvious. Andreas Floros, professor of computer science, rector of the Ionian University and chairman of the Synod of Rectors, reported this to K.who at the next meeting in March will address ChatGPT issues for the first time.

“This is software that can create dialogues in response to short or long texts given to it by users. That is, it is a robot that we feed information through the Internet, and when we ask something, it tries to generate answers based on what it has learned. I would liken him to a student who read countless texts, but did not understand them, but repeated them like a parrot.” Associate Professor of the Department of Communication, Media and Culture Panteio Kostas Karpuzis, tells “K”.

“Such systems are designed to answer questions in a natural, dialectical, interactive way, to reproduce speech, and not just to suggest sources of information. They don’t just provide search results and access to content. They collect, process, synthesize data and generate responses by associating patterns and information. They are constantly trained to work with huge and enriching sets of textual data.” adds “K” Professor of the Department of Engineering Information and Communication Systems of the University of the Aegean Lilian Mitrow.

The use of ChatGPT for writing student assignments or teleexam answers, and thus the problems it creates for educational institutions, are obvious. “The sciences and arts have always used technological tools. For example, writers have moved from the pen to the typewriter, computer, online search tools, and automatic translators. The same thing is happening now, but in a different way: creative (or generative) neural networks create “generational pens” and, as they are called, “generational brushes” that, in capable hands, can produce new and never-before-seen art forms. and science, even poetry.

The difference that ChatGPT made was to build texts that have fluidity and perhaps argumentation, so they indicate elements of natural intelligence,” explains Ioannis Pitas, computer science professor at AUTH, in whose department they were recently examined, as if he was a student. ChatGPT in digital processing. It received a score of 8 out of 10. The information from “K” says that in many university departments in the country, professors study the “bad” use of ChatGPT by students. As Mr. Pitas notes, the most vulnerable areas are theoretical sciences, economics and medicine, while ChatGPT struggles with questions that contain forms, that is, with architecture and IT departments.

ChatGPT will not make a difference for live exams in a classroom without mobile phones, but can be used on remote exams or assignments. This is why it is learned that students also take oral examinations on the topic of the papers they submit. “If you’re looking for depth in matters, ChatGPT fights. We try to adapt the exercises so that ChatGPT cannot answer them, while before exams we ask ChatGPT questions, and after exams we check students’ written work. Basically, we “test” exercises on ChatGPT before we give them to them,” adds Dimitris Fotakis, professor at NTUA’s School of Electrical and Computer Engineering.

On the other hand, as Mr. Karpuzis notes, ChatGPT has weaknesses: “For example, at this stage he will not be able to correctly solve a physics problem in elementary or high school, because he did not understand the relationship between force and acceleration or energy and speed body – he tries to create a text that is as similar as possible to what he studied.

“ChatGPT formalizes an evolution that has taken place in the educational process: the completion of a written test as an effective means of assessment. We see this at the graduate level, with dissertations becoming increasingly non-original, often products collected from the Internet. Now the examinee’s attitude towards any work presented as his own is revealed only at the presentation stage,” emphasizes Giorgos Dellis, professor of public law at the Athens School of Law. “Because one of the main applications of generative AI systems is text generation, ChatGPT has been viewed as a Trojan horse for school and university copying and plagiarism from the beginning. Although the production of texts in Greek will never be as efficient as in Western languages, simply because there is no commercial interest in training such a system for the required amount of texts, we cannot ignore this risk, especially at a time when systems that evaluate the likelihood that the text is written according to such an algorithm, they will always be one step behind potential imitators, ”says Mr. Karpuzis.

“In fact, this is not so much plagiarism as writing an article by a “third party”, which is also an unacceptable form of fraud. The software in question does not copy the sources directly, but recomposes – reworks the existing material. Thus, the process is more like a contribution from a third party that is not an individual – it should be noted that writing articles for a fee by third-party authors has unfortunately been the answer in universities for many years – but the application of artificial intelligence, which gives an immediate result and free of charge,” emphasizes Antonis Karabatzos, professor of civil law at the Athens School of Law. He highlights the dangers, saying that “this phenomenon is unprecedented and threatens the fundamental principles of academic ethics as well as original intellectual creativity and critical thinking. And while it can sometimes be easy to identify the use of software like ChatGPT, it’s not at all easy to prove because the software gives a unique answer to every question.”

“Cosmogenic change in cognitive tools should not cause panic. Haven’t we moved into the era of podcasts instead of the traditional academic newspaper? As for the written word, let it be a few scientific creations that claim to be unique (doctoral studies, monographs). This is where digital technologies come in handy because they detect plagiarism, regardless of the human or mechanical origin of the reproduced text,” notes Mr. Dellis. “But perhaps this is also a good opportunity to reflect on what exactly we require from our students and students in terms of knowledge and, above all, critical abilities: can the answers we ask for in the exam give … a well-read student , but read and do not understand or have not mastered the material, then ChatGPT will indeed receive a rating very close to excellent, ”adds Mr. Karpuzis.

As Mr. Pitas emphasizes, “The real danger is that AI projects can fool many people, even experts, and worse, scare people. A frightened society can only be self-destructive. We saw this recently with the coronavirus. Fortunately, we cannot stop scientific progress. Those who did it, i.e. the Luddites failed miserably. What we can do; To educate the world so that we have citizens and scientists with critical thinking, generalization, imagination, emotion and knowledge. But this needs to be done now and at the global level. Otherwise, humanity is threatened with an internal collapse if it is unable to cope with the production and transfer of knowledge. This happened in the past, for other reasons, at the beginning of the Middle Ages. But now we know better. Having successfully dealt with the nuclear risk, we can do the same with the current risks of artificial intelligence.”

“I will include the app in my course as a tool”

The decision to use ChatGPT next spring semester was made by Rebecca Paidi, Associate Professor at the Department of International and European Studies at the University of Macedonia.

Teachers from the Faculty of Communications, Media and Culture at Panteillon University will do the same. Is this a heretical position at a time when the university community is looking for ways to overcome the challenges that ChatGPT will create in the educational process? “ChatGPT is not the enemy. I will use it as an active teaching aid in the Foreign Policy Analysis course. In individual and group exercises in the classroom, students will use ChatGPT as a partner, drawing ideas from his answers, evaluating answers, or looking for mistakes he makes,” notes Ms. Paidi.

She believes that the technology will help identify text coming from artificial intelligence tools in the future. “However, daemonizing ChatGPT is useless. Our students will use it anyway. Better to show them ways to use it constructively.

The exercises for critical analysis of ChatGPT results, for example, are varied. And this is because he combines knowledge with critical skills and cultivates the belief that we cannot perceive information on the Internet uncritically, without a knowledge filter. Comparing an essay written by a student with an essay offered by ChatGPT on the same topic can reveal strengths and weaknesses and provide alternative points of view. “ChatGPT can prove to be a useful classroom partner at a time when teaching and learning are facing revolutionary changes,” she notes.

On the other hand, in addition to using ChatGPT for content creation, artificial intelligence is also an extremely interesting subject of research not only for computer scientists, but also for the humanities. As Kostas Karpuzis, Associate Professor of Communications, Media and Culture at Panteio, told K, “In my department, we will use ChatGPT and related technologies in two courses over the next semester, creating original content that will be used during the development of digital games (scenario , visuals, sound effects and music), but also learning how these technologies reproduce the gender and race stereotypes contained in the data they are trained on: just ask ChatGPT to write you a text about a hypothetical doctor and count how many texts, which he creates, a name referring to a woman or an immigrant is used.