In recent years, artificial intelligence has become a familiar phenomenon and has firmly entered everyday life. A bot can answer almost any question, explain a difficult topic, paint a portrait, create a music video, and even compose a song or a poem. Education has not remained untouched by this trend.
In fact, neural networks first appeared in the mid-20th century and have been evolving for several decades.
-As far back as 1994, I wrote my diploma thesis on “Controlling the Movement of a Transport Robot Using a Neural Network.” In 1997, at the Saint Petersburg Institute for Informatics and Automation of the Russian Academy of Sciences, we were working on issues related to speech recognition, developing what is now called language models. Unfortunately, at that time, the level of computer technology did not allow us to achieve the results we see today - says Alexey Makhovikov, Director of the Institute of Basic Engineering Education at Empress Catherine II Saint Petersburg Mining University.
Today, 70% of students in Russian universities actively use artificial intelligence in their studies — for information searches, preparation for seminars, and exam revision. This is confirmed by a survey conducted by the IT company MWS AI. Moreover, the overwhelming majority of respondents (79%) consider the use of neural networks during exams and in the preparation of examination papers (such as term papers, theses, etc.) to be acceptable.
The scope of applications of computer intelligence in education is indeed extensive. Students can ask a program (most often ChatGPT, Gemini, or YandexGPT) to generate ideas and outlines, assist in formatting, compile bibliographic lists, and formulate goals and objectives. There are also less “ethical” requests: to write one or several sections of a paper (an introduction, a conclusion, or a theoretical chapter), to paraphrase one’s own or someone else’s text, or even to create an entire work from scratch — whether an essay or even a term paper.
It is hardly surprising that with the growing popularity of new technologies, situations of varying degrees of scandal have begun to “surface.” In Russia, the first high-profile case occurred in January 2023, when RGGU student Alexander Zhadan successfully defended his thesis, which had been generated for him by ChatGPT. The young man himself admitted to the “affair” by posting about it on social media. This precedent sparked a storm of discussion and literally shook the academic community. Experts described it as a challenge to the entire system of Russian higher education. RGGU itself did not revoke the graduate’s diploma; moreover, the university did not introduce a blanket ban on the use of neural networks, though it did call for restricting access to ChatGPT within educational institutions.
One thing is clear — discussions about the appropriateness and role of artificial intelligence in higher education will not end anytime soon.
Anton Volkov, a graduate of the Kikot Moscow University of the Ministry of Internal Affairs of Russia, recalls that he never used the “services” of neural networks himself: it took too much time to verify the output. According to him, the system often made many logical mistakes and strange conclusions. Incidentally, what his classmates considered a tedious and boring task (searching for information, drawing up work plans) turned out to be an important part of learning for Anton.
- I am a future lawyer, and for me, meticulous fact-checking and the systematization of information are crucial. A neural network often “hallucinates,” producing a plausible version that is in fact false. Essentially, this means a complete failure of the work, which discredits me as its author. I cannot allow that,, - Anton says.
- Mindless copying has always existed: ready-made homework, custom-written essays and term papers. In school, some copied from classmates. In my view, the problem is not artificial intelligence itself, but rather explaining to young people that this tool should serve the person, not replace human intelligence, - says Anton’s mother, Tatyana Volkova.
We attempted to determine whether artificial intelligence is perceived as a problem within universities themselves, and how universities can become a platform for responsible and effective “collaboration” between humans and new technologies.
- The overwhelming majority of universities and faculty members are concerned about the mass spread of AI tools; this is comparable to the plagiarism problem of 10–15 years ago. It is a challenge to the entire education system, which is now forced to reconsider its approaches to assessing knowledge, — says Evgeny Mazakov, Head of the Department of Information Systems and Computer Engineering at Empress Catherine II Saint Petersburg Mining University.
Most experts are in favor of neural networks, but on the condition that their use is transparent and openly declared.
- Should we forbid people to switch from horses to motorcycles? If we are talking about horse racing, then perhaps yes. But if the task is to deliver something as quickly as possible, then prohibition makes no sense. The same reasoning should be applied to artificial intelligence, — draws an analogy Olesya Koltsova, Professor at the Department of Sociology at HSE University — Saint Petersburg and Head of the Laboratory for Social and Cognitive Informatics.
Evgeny Mazakov agrees with this position. He believes that the issue is not so much the use of AI itself, but rather its uncontrolled and unethical application. Irresponsible use of neural networks carries many negative consequences.
- First, the educational objective is lost. A student who does not write their own work does not acquire key competencies — the ability to analyze information, think critically, structure thoughts, and formulate conclusions. Second, there is the risk of unfair assessment, since the grade is given not for the student’s knowledge and effort, but for the output of an algorithm. Third, there is a significant degree of unreliability. AI often generates material that appears plausible but is factually incorrect or entirely fabricated, especially in highly specialized areas (the so-called “hallucinations”), — the expert explains.
For now, most universities are only in the process of developing official policies and regulations on the use of artificial intelligence. Most often, the issue is regulated at the level of departments or even individual instructors. Some set rules in the methodological guidelines for specific courses. However, a general principle is already in effect: any use of AI must be transparent. In other words, a student must honestly indicate what exact function the bot performed. For example, if a neural network was used to generate an idea or a plan, this should be specified in a footnote or an appendix.
- Passing off generated text as one’s own is considered academic misconduct. The consequences are equated with those for cheating or plagiarism and depend on the internal rules of the university and the seriousness of the violation: from a failing grade and prohibition from defending the work to disciplinary action, such as a reprimand or even expulsion, - Mazakov notes.
What are universities doing right now, and what tools are they using to control the situation? Existing anti-plagiarism software is not always effective, so other methods are being employed.
At Empress Catherine II Saint Petersburg Mining University, instructors are increasingly changing the format of assignments, preferring oral forms: project defenses, colloquia, and interviews based on written work. This makes it easier to quickly determine whether a student truly understands the material.
- Practice-oriented assignments also work well: tasks based on personal experience, data from a specific enterprise, or a unique experiment or study that cannot be generated. Likewise, creative and analytical tasks — essays that require a personal stance, a critical analysis of a case, or an examination of recent news and events not yet reflected in AI datasets, - says Evgeny Mazakov.
In most universities, there is no outright ban on the use of AI by students. However, active work is underway to develop methods for detecting undeclared use, particularly in written texts. Individual tests and group assignments carried out in class are also employed.
- For example, in my course “Design and Methods of Data Collection in Sociological Research,” I check in an individual test how well a student can formulate questions in a questionnaire for respondents. They must be able to do this independently, just as they must understand how many conditions their experiment should have in order to test their hypotheses. Therefore, the use of AI in this case is prohibited. As a temporary solution, I have moved this test to paper. However, they also have a large team assignment on developing an experimental design and questionnaire for given research hypotheses, and in this case, if students are able to harness artificial intelligence to successfully complete the task, then that is to their credit. At the same time, if I see something incoherent in the work because students did not think about what to ask AI or were unable to verify its response, they will have to redo the assignment, - explains Olesya Koltsova.
At the same time, the “use” of AI has certain advantages in terms of enhancing education. At HSE, it is noted that students are quickly and effectively acquiring such a skill as the ability to formulate precise prompts (queries).
According to experts, the question is not whether to allow or forbid neural networks. It is entirely possible that in the future, bots will become as commonplace a tool as a calculator for an engineer or a text editor for a writer. The future of AI in Russian universities lies not in prohibition, but in integration and adaptation.
What changes are likely to take place in universities in the foreseeable future with the rise of neural networks:
- A shift in approaches to education. Universities will place less emphasis on the simple reproduction of information (which AI performs flawlessly) and increasingly focus on critical analysis, verification, interpretation, and the application of knowledge in non-standard situations.
- New specializations. Courses and programs devoted to prompt engineering, AI ethics, and the development and implementation of AI solutions in professional fields are already emerging. Their number will only continue to grow.
- At present, the use of neural networks without acknowledgment poses a significant risk. Strategically, however, ignoring AI is not an option. The future lies in understanding its capabilities and limitations, and in learning to use it ethically and effectively to enhance one’s own intellectual abilities, rather than to replace them, - summarizes Evgeny Mazakov.
- The main task of educators at this stage is to determine what we still expect students to be able to do independently, as before; what is no longer necessary to require of them; and what, on the contrary, they must now learn to do for the first time — and specifically in interaction with AI. Here, of course, the role of the academic community together with state authorities is crucial. I believe it remains an open question for discussion: what exactly can be delegated to AI, and what should not, - adds Olesya Koltsova.
At Russia’s oldest technical university, as part of the Pilot Project, a compulsory elective course entitled “Intelligent Software” has been introduced for second-year students to familiarize them with modern artificial intelligence systems. The course teaches students to use existing technologies for the effective solution of engineering problems, while also addressing the ethical aspects of the issue.
- In training IT specialists in the senior years, we also take into account the modern realities associated with the development of AI. We teach students to use available tools competently, delegating routine tasks to AI while retaining full control over project implementation. Six years of study within the framework of the Pilot Project enables us to prepare fully qualified information technology engineers who will be valued in the labor market, - concludes Alexey Makhovikov.
