Использование ИИ российскими учеными и преподавателями: результаты исследования

Researchers from ITMO University's Center for Science Communication, Yandex Education, and the Yandex Cloud Center for Society Technologies surveyed faculty at 16 Russian universities – specifically, those who both teach and conduct research. The results were telling: two out of three respondents use AI tools in their work on a regular basis.

This is no longer a story of “trying it once”; it's about a fully-formed habit. Let's explore what people are actually doing with AI, where it truly helps, and where it's still stalling.

Применение ИИ в работе ученых и преподавателей

Where AI Has Already Taken Root

The most popular task is working with text. Academics and educators actively use AI for editing and improving phrasing, translating materials, writing abstracts, and structuring large volumes of information. Simply put, everything that used to be time-consuming but didn't require deep expertise is now being delegated to machines.

In second place is literature search and analysis. When quick familiarization with a topic or finding intersections between different studies is needed, AI significantly speeds up the process. This is especially valuable for those working at the crossroads of several fields of knowledge.

Educators, in turn, have found a use for AI in preparing course materials: creating lesson plans, generating examples, and adapting explanations for different audience levels. What used to take an entire evening now takes much less time.

Проблемы и ограничения использования ИИ в академической среде

It's Not All Smooth Sailing

However, attitudes toward AI in the academic community are far from uniform. A sense of professional caution is noticeable among the survey participants – and this is, perhaps, the right stance to take.

The main complaint is reliability. AI can confidently present inaccurate information, invent non-existent sources, or mix up facts. For academic work, where every statement requires verification, this is a serious drawback. Most respondents note that AI-generated results must be double-checked, which adds work rather than eliminating it.

There is also a more subtle question: where does assistance end and substitution begin? This is particularly acute in teaching. If a student submits AI-written work and the instructor uses AI to grade it, what is happening to the learning process? This question remains open, and there is no single answer within the academic community.

Отличия в использовании ИИ молодым и опытным поколением

The Young Are More Active, The Experienced More Cautious

The study noted an expected but important divide: younger faculty and graduate students are adopting AI tools more quickly and willingly. They grew up in an environment where technology is constantly changing and see AI as just another work tool – much like a search engine or a spreadsheet.

The older generation of researchers is more reserved. Here, it's not about age itself, but rather established professional habits and a higher bar for source requirements. Someone who has spent decades developing a methodology for working with information is in no hurry to change their approach for the sake of speed.

This isn't a conflict – it's more like different entry points into the same transformation.

Этические вопросы и авторство в эпоху использования ИИ

Ethics and Authorship: Questions Without Answers

The study also highlights the topic of ethics. When AI helps write an article, who is the author? Should this be disclosed? How does this align with the requirements of journals and conferences?

As of now, there are no uniform standards. Some publications require disclosure of AI use, others prohibit it altogether, and still others do not regulate the issue at all. The academic community is just beginning to develop these norms – and judging by the pace at which AI is entering daily practice, there isn't much time left.

A broader concern was also voiced among the participants: won't the mass adoption of AI lead to research becoming homogeneous? Will originality of thought, unconventional hypotheses, and the authorial voice of the scientist suffer?

For now, these are more intuitive fears than a documented problem. But they show that people are thinking not only about convenience but also about the longer-term consequences.

What This Says About the Industry as a Whole

The academic world is not the fastest to adopt new tools. If even here two-thirds of specialists are already working with AI regularly, it indicates that the technology's penetration has moved beyond the IT sector and startup culture.

However, the nature of its use in science and education differs from that in, say, marketing or customer service. Here, the standards for accuracy are higher, professional responsibility for the outcome is greater, and the question is far more acute: when does a tool help you think, and when does it start thinking for you?

Judging by the study's results, Russian academics and educators are, for the most part, keeping this question in mind. They use AI pragmatically – where it genuinely saves time – and remain skeptical where accuracy and authorship are at stake. This is, perhaps, the most sensible balance at this stage.