RU

Keyword: «neural network training»

Full text Read online
In recent years, the processes of education informatization are characterized by significant qualitative changes associated with artificial intelligence (AI) technologies. We are primarily considering neural networks that process graphic and text information. Their rapid development has already had a significant impact on the course of the educational process and will undoubtedly have an even greater impact in the near future. In this context, the authors present a study aimed at identifying the pedagogical potential of neural networks for image generation in relation to higher education. The objectives of the study are: to identify the specific aspects of the current level of neural networks development in relation to possible pedagogical applications; identifying the capabilities of neural networks for generating images that can be used to solve problems relevant to modern pedagogy; finding the main directions of neural networks application for image generation in the pedagogical process; analysis of the readiness degree of students, as subjects of the educational process, to work with this technology; identifying the level of students’ awareness of neural networks work, attitudes towards their use, identifying problems, identifying the most accessible and convenient neural networks for use in teaching, justifying the need to include work with such applications in the content of education. Based on a comprehensive review of domestic and foreign research works, a conclusion is made about the significant pedagogical potential of AI image generators and their leading pedagogical application – the generation of educational content. The capabilities of the main currently available AI image generators are analyzed – Kandinsky 2.1, Lexica art, Masterpiece, Dream by Wombo, Craiyon and Playground AI. Their functionality, capabilities and limitations are considered. The results of a survey of students who have experience of working with AI image generators are analyzed. Data is provided on what neural networks the students worked with, what the main directions of their use are, according to the respondents, what advantages and disadvantages neural networks have, and what difficulties the respondents encountered during the work. What seems significant is the conclusion that, according to the majority of respondents, artificial intelligence systems, and in particular AI image generators are preferable to use as a source of ideas for their own learning activities. The theoretical significance of the study is in the fact that the pedagogical potential of AI image generators in relation to higher education has been identified and a key area of their application has been found – the generation of educational content. The practical significance of the study is in identifying a set of factors to increase the efficiency of using AI image generators in higher education, including training of potential users (both students and teachers) in the interface, the basic principles of neural networks operation, as well as procedures for interacting with them. The main “danger area” in the use of artificial intelligence systems by students has also been identified – the desire to use them as a “source of ideas” and a mechanism for relieving this risk has been proposed, consisting of introducing competitive elements into the process of interaction between the student and the AI system.