"Human or artificial intelligence?": an empirical study of respondents' declared preferences in digital graphics

Solomentseva Svetlana Borisovna ORCID: 0000-0002-2442-1235 Associate Professor; Department of Design, Art Education and Technology; Federal State Budgetary Educational Institution of Higher Education 'Bunin Yelets State University' 399770, Russia, Yelets region, Kommunarov str., 28 ss00001@mail.ru

Introduction

The relevance of the research topic is determined by the fact that currently technologies based on the use of artificial intelligence (AI) capabilities are being implemented in almost all spheres of human life. At the initial stages of this process, the main task was to optimize routine production algorithms, but recently there has been a tendency to integrate AI into areas such as culture and art. Innovative visual technologies are developing rapidly and are already competing with designers and artists.

A number of Russian and foreign scientists have attempted to consider generative creativity in terms of uniqueness, aesthetic value, philosophical and ethical aspects. Researchers from Samara State Technical University (Russia) note that attempts to combine art and technology were made by Leonardo da Vinci in the Renaissance and propose to perceive neural networks as the next stage in the development of society, the rational use of which will reveal new facets of human creativity ^[1]. Colleagues from Saratov National Research State University named after N. G. Chernyshevsky (Russia) have substantiated the socio-cultural prerequisites for the emergence of AI, the high potential of its influence on thinking and artistic expression of personality ^[2]. The work of scientists from Simon Fraser University (Canada) is devoted to the study of the philosophical aspect of the introduction of generative technologies into visual practices. Intelligent systems are positioned as a means for creating works based on dynamic interaction with humans and a new cognitive structure of creativity ^[3].

The study, carried out on the basis of the Moscow Pedagogical State University (Russia), shows the changes in the modern paradigm of art in the context of the integration of science into the creative process. AI is considered as an auxiliary tool that provides a wide range of personality-oriented thematic materials so that people can more effectively realize their potential in creating new forms of self-expression ^[4]. Scientists from the University of Tampere (Finland) studied the attitude of a group of respondents to the introduction of AI in the field of creativity from the perspective of psychological comfort. The results obtained indicate an ambiguous attitude of the survey participants towards neural network art. Despite the generally positive reviews, many found the generated images strange and even frightening, due to the lack of positive experience using such tools ^[5].

The work of researchers at the University of Science and Technology of Macau (China) is devoted to the analysis of the impact of AI on graphic design. Four basic concepts have been identified, covering a wide range of possible uses of these technologies, ranging from purely technical aspects to the emotional impact on viewers. Neural networks, when used consciously, can not only increase the efficiency of the project process, but also stimulate the creative thinking of designers interacting with them ^[6]. Colleagues from the Moscow State Psychological and Pedagogical University (Russia) highly appreciated the potential of AI application in the field of video art ^[7]. Researchers at the Industrial Design Center (China) analyzed the potential impact of generative technologies on the industrial design industry. It is rational to use AI in specialized clusters, where high speed and mathematical accuracy are required, and the conceptual component should remain the prerogative of the creative personality ^[8].

Scientists from Macau City University (China) have studied neural network technologies as a factor in increasing the effectiveness of visual communication of multimedia projects. At each stage of creating layouts, the AI's work was evaluated by designers in terms of consistency with the concept, harmony of composition, and readability of typography, and adjusted if necessary. Statistical analysis of the data obtained using the Likert scale has shown that illustrative materials created through interactive interaction between humans and neural networks are more preferable for viewers than layouts made using traditional methods. This opens up new opportunities for the mutual penetration of art and technology, improving the quality of visual projects, and developing new areas of design activity ^[9].

Researchers at the University of Cambridge (UK) are confident that AI-based tools will be widely used in various spheres of society, including art, and suggest that educational institutions now introduce interdisciplinary courses in this field into their curricula ^[10]. Colleagues from the School of Art of Southeastern University (China) advocate even more significant changes aimed at transforming the model of art education, taking into account the capabilities of new neural network technologies ^[11].

However, not all representatives of the scientific community are so optimistic. Scientists at Ruhr University (Germany) pay tribute to the opportunities that AI provides to modern society, but consider the spread of these technologies as a threat to future generations, since a person equipped with such tools may stop thinking, analyzing and creating. As a way out of this situation, it is recommended to focus on the development of critical thinking and the independent work of the creators of design projects. It is also proposed to develop norms regulating the rational number of generated elements in artistic works ^[12]. Researchers at the Higher School of Economics (Russia) note that AI is based on controlled algorithms that create predictable and fairly standardized images. The evaluation of these projects shows the absence of a deep subtext and moral component, which is inherent in the works created by artists, and makes them works of art that remain relevant for centuries ^[13].

To objectively assess the potential of generative technologies and their ability to compete with copyrighted works, it is necessary to consider creative projects from a more formal point of view. The purpose of our study is to empirically identify respondents' declared preferences in digital graphics between works created by humans and artificial intelligence. The study and analysis of the opinions of representatives of modern society will allow us to determine to what extent neural networks will influence the trends in the development of the field of artistic creativity.

Materials and methods

This article reflects the results of a study conducted at the Department of Design, Art Education and Technology of I. A. Bunin Yelets State University during 2024. The preparatory stage was quite laborious and consisted of two stages. On the first of them, artists and designers created original images in various styles of digital graphics based on a detailed description. Various software tools were used during the work.: Adobe Photoshop, Krita, Procreate, HiPant, etc. The purpose of the second stage was to generate illustrations based on the same description using neural networks: Midjourney, DALL-E, Kandinsky, Stable Diffusion, Masterpiece, etc. The leading specialists of the department analyzed and selected the options that were included in the test questionnaires for empirical research. Two impersonal versions are presented for each description: the author's image, created by a person and developed using generative technologies, examples are shown in Figures 1, 2.

Fig. 1. Examples of images included in the test questionnaires: a - the work of Ksenia Povolyaeva, b - the result of generation in a neural network

Fig. 2. Examples of images included in the test questionnaires: a - the work of Maria Kuznetsova, b - the result of generation in a neural network

At the main stage of the study, a target group was formed, which included sixty-two respondents from among teachers and students of creative fields of study at the I. A. Bunin YSU, detailed information about which is presented in Table 1 and the diagram in Figure 3.

Table 1

Information about the target group of respondents

Fig. 3. Chart of the percentage ratio of categories of respondents

Our opinion coincides with the statement of scientists from the Center for Industrial Design (China) that the difference in the profile of basic education of the survey participants affects the correctness of the final indicators ^[8]. Therefore, the inclusion of people with artistic experience in the target group determines a fairly high degree of reliability of the results obtained, since they are based on objective judgments and opinions of respondents about the topic under consideration. To collect empirical data, the method of declared preferences was used, implemented in a remote format using Google Forms, which allowed the study participants to independently determine the required duration of the testing procedure, time mode and comfortable location. During the performance analysis stage, the methods of quantitative statistical analysis implemented in the Microsoft Excel program were used.

Results and discussion

The specifics of evaluating works of visual art are such that it is based not only on objective data, but also on subjective human perception. Due to the significant increase in the number of specialized websites and groups on social networks aimed at connoisseurs of computer creativity, art criticism often "gives way to popular vote" ^[14]. We support the point of view of colleagues from the Russian University of Technology that, to date, there is no theoretical basis for a critical analysis and an unbiased assessment of works performed by means of digital graphics ^[15]. In our study, we used the method of declared preferences, which is based on the personal opinion of the members of the target group of respondents about the quality of works created by humans and AI ^[16]. The results are presented in Table 2 and the diagram in Figure 4.

Table 2

Data on respondents' declared preferences

* Total number of choices = number of respondents × number of image options created according to different descriptions. In the course of our experiment, each of the 62 respondents was tasked with determining their preferred option from 10 pairs of images created according to different descriptions of the author and generated images.

Fig. 4. Diagram of respondents' declared preferences

The results obtained during the study show that the students of secondary vocational education earned the highest degree of approval for the work created using AI, they scored 52.8% (the answer is "yes" – 11.4%, the answer is "rather yes" – 41.4%). However, digital author's images are also quite popular, their figure is 47.2% (the answer is "yes" – 19.3%, the answer is "rather yes" – 27.9%). These data clearly illustrate the processes taking place in the modern art space, where generative services are becoming more diverse, accessible and intuitive for people with a high level of computer technology proficiency. However, neural networks, in addition to their wide capabilities and apparent ease of use, carry certain risks and threats. Our point of view coincides with the opinion of researchers from the International Association for Intercultural Dialogue and Geostrategic Research (Switzerland) that representatives of creative professions may become addicted to the uncontrolled use of AI, which can potentially suppress their own creativity ^[17].

Undergraduate students distributed their announced preferences as follows: human–made works received 52.9% (the answer is "yes" – 21.8%, the answer is "rather yes" – 31.1%), generative technology products scored 47.1% (the answer is "yes" - 21.8%, the answer is "rather yes" – 31.1%). According to this category of respondents, the current level of development of AI technologies is such that the difference between generated images and author's projects is no longer so obvious. This is consistent with the opinion of scientists from Finland, Turkey and China, who conducted a joint study to analyze the degree of influence of neural networks on modern visual art and propose using AI to reduce the time required to create high-quality illustrative materials and visual components for event design ^[18].

Students enrolled in master's degree programs demonstrated their commitment to digital author projects with a score of 57.5% (the answer is "yes" – 26.7%, the answer is "rather yes" – 30.8%), while illustrations developed by AI received 42.5% of the votes (the answer is "yes" – 17.5%, the answer is "rather yes" – 25.0%). The survey participants note that the images created in neural networks are perceived by viewers as more innovative, but devoid of authenticity, which leads to a certain degree of devaluation of both specific works and the artist's work as a whole. The same conclusion was reached by scientists from the University of the Bundeswehr in Munich (Germany), who examined various aspects of the use of elements of generative art during the creation of works of art. The use of AI capabilities will be the most rational when creating commercial visual projects, the duration of which is relatively short and the investment of significant designer labor is unreasonable ^[19].

A significant part of the declared preferences of the faculty respondents were given to human-made images – 76.2% (the answer is "yes" – 7.5%, the answer is "rather yes" – 68.7%) and only 23.8% (the answer is "yes" – 3.8%, the answer is "rather yes" – 20.0%) got projects drawn by AI. According to teachers, the widespread introduction of neural network technologies, in addition to the positive aspects of freeing artists from routine procedures, carries significant risks. The creation of images takes place on the basis of average algorithms, which can lead to the loss of cultural and personal identification of a modern person. A similar point of view is shared by colleagues from Saratov National Research State University named after N. G. Chernyshevsky. A solution to this problem may be to control the percentage of the author's and generated parts of the works, as well as to address the spiritual and emotional component of art ^[2].

The overall results of the study for all categories of respondents indicate that currently human-made creative works are still more preferable for members of the target group, in general they received 55.5% (the answer is "yes" – 20.3%, the answer is "rather yes" – 35.2%). However, images developed by neural networks compete quite seriously with an indicator of 44.5% (the answer is "yes" – 13.5%, the answer is "rather yes" – 31.0%). The data obtained are consistent with the statement of scientists from Osh State University (Kyrgyzstan) that AI already has a significant impact on the creative environment ^[20]. It seems logical to us to divide the creative process into logically justified stages. The basis for the development of art projects should be the author's conceptual images, and the capabilities of AI can be implemented at technological stages that require monotonous and lengthy work. The generative direction in digital graphics has great potential and wide possibilities, but to create highly artistic works, human thinking is necessary, capable of transforming emotions and feelings into visual images.

Conclusions

The scientific novelty of the results obtained lies in the fact that the study identified the respondents' declared preferences between digital graphic works created by humans and artificial intelligence. This is relevant at the present time, because society is on the verge of large-scale changes, when the basic concepts of art and creativity are undergoing transformation.

The focus group included 62 representatives of the faculty of I. A. Bunin Yelets State University and the student community with experience in conducting artistic activities. This fact indicates a sufficient level of reliability of the research results, since the respondents' opinion is based on existing knowledge, skills and abilities. What is especially important, since today there are no specific parameters for evaluating digital art, a classical approach based on emotional perception and human experience is used.

The empirical results show that the highest degree of approval of respondents was given to human–made works, they scored 55.5% (the answer is "yes" - 20.3%, the answer is "rather yes" – 35.2%). Digital images generated by AI are also very popular, with an index of 44.5% (the answer is "yes" – 13.5%, the answer is "rather yes" – 31.0%). However, there are significant differences in the opinions of the survey participants in different categories. Students of secondary vocational education, aged 17-20, who are more open to innovation in all fields, including art, preferred the creation of neural networks – 52.8% (the answer is "yes" – 11.4%, the answer is "rather yes" – 41.4%). Whereas teachers aged 29-57 have a well–established opinion about the necessary qualitative parameters of artistic works, their sensory perception and semantic connotation, therefore their declared preferences for generated images are only 23.8% (the answer is "yes" - 3.8%, the answer is "rather yes" – 20.0%).

One of the main goals of educational institutions is to train competitive graduates with socially demanded skills, therefore, the widespread use of AI makes it necessary to introduce modules focused on the study of generative technologies into the educational process. The results of the research are integrated into the master's degree program in the field of preparation 04/24/2011, the profile "Creative design and fashion illustration". It is assumed that content analysis, interpretation and evaluation of works created using AI will enhance the skills of creative expression and the level of professional skills of designers of the future.

In the course of the study, we came to conclusions similar to the opinion of colleagues from Magnitogorsk State Technical University named after G. I. Nosov (Russia) ^[21] and scientists from the National Center for Contemporary Arts (Belarus) ^[22], that it is necessary to take into account not only technical aspects, but also the ethical foundations of the introduction of AI in visual practices. Paying tribute to the functionality provided by neural networks, we suggest classifying them as a tool that helps artists translate original ideas into creative graphic projects. The harmonious combination of the traditional approach and the advantages of AI-based digital technologies is a guarantee of creating a successful art product. The vector of development of modern art clearly demonstrates new forms of interaction between participants in the creative process and allows us to rethink the algorithm for creating artistic works.