AI tools for science: basic classification, strengths, weaknesses, learners’ opinions

Importance. Training personnel for careers in science and the economy requires modern research competencies in the scientific and technological sphere, including mastery of AI technologies. This paper aims to develop a basic classification of AI tools applicable to undergraduate, graduate and postgraduate students of the humanities, and to analyse students' subjective opinions about the effectiveness, strengths and weaknesses of using AI in science.

Research Methods. The following scientific methods are employed: analysis of relevant literature; a training experiment; a formalised questionnaire; and statistical methods.

Results and Discussion. The research revealed that undergraduate students utilise AI to structure information (73.9 %), write conclusions (78.3 %), write a literature review (60.9 %), and generate ideas (52.2 %). Masters and PhD students use AI to design articles and reference lists (73.9 %). In the survey, respondents identified ChatGPT (  = 8.5 and 8.2 points), DeepSeek (  = 8.2 and 7.7 points) and Chatpdf ( = 7 and 7.7 points) as the most effective resources. Master's and PhD students demonstrated a heightened level of critical thinking when evaluating the strengths and weaknesses of AI tools. They were more likely to identify potential limitations.

Conclusion. The differences between Masters’ degree Students/Post-Graduate Students in the choice of AI resources and in the assessment of their advantages and disadvantages are due to the different levels of their research competence and the degree of readiness for independent scientific activity. The application of AI can facilitate students in solving a number of tasks, but only qualified teachers are able to supervise their research and inform them of the correct and incorrect ways to use AI in science.