ARTIFICIAL INTELLIGENCE AND SOFTWARE DEVELOPMENT IN SLOVENIA: ADOPTION, CHALLENGES, AND OPPORTUNITIES

УДК 004.8

ABSTRACT

Artificial intelligence (AI) is already transforming the software development sphere. However, it remains a controversial topic. AI offers opportunities for enhanced productivity, creativity, and efficiency. At the same time, it is raising concerns about security, ethics, and workforce implications. This study comprehensively investigates the integration of AI in Slovenian software development across various stages and organizational levels. The study is exploratory and qualitative. It consists of two parts, combining a scoping literature review with structured interviews of professionals. The results identify the utilization rate, purposes, most popular tools, and the impact of age and organizational role on utilization. Notable discrepancies between global and Slovenian practices suggest the need for further research into local integration strategies. The findings provide valuable insights for strategic AI implementation in both Slovenian and international software development contexts, particularly in similar countries in Central and Eastern Europe. Additionally, they can serve as a good starting point for further research on the same or related topics.

АННОТАЦИЯ

Искусственный интеллект (ИИ) уже трансформирует сферу разработки программного обеспечения. Однако он остается спорной темой. ИИ открывает возможности для повышения производительности, креативности и эффективности. В то же время он вызывает опасения по поводу безопасности, этики и последствий для рабочей силы. В данном исследовании всесторонне изучается интеграция ИИ в словенскую разработку программного обеспечения на различных этапах и организационных уровнях. Исследование носит поисковый и качественный характер. Оно состоит из двух частей, сочетающих обзор литературы со структурированными интервью с профессионалами. Результаты определяют уровень использования, цели, наиболее популярные инструменты, а также влияние возраста и организационной роли на использование. Заметные расхождения между мировой и словенской практикой указывают на необходимость дальнейших исследований локальных стратегий интеграции. Полученные результаты предоставляют ценные сведения для стратегического внедрения ИИ как в словенском, так и в международном контексте разработки программного обеспечения, особенно в подобных странах Центральной и Восточной Европы. Кроме того, они могут служить хорошей отправной точкой для дальнейших исследований по данной теме.

Keywords: artificial intelligence; software; software development; Slovenia.

Ключевые слова: искусственный интеллект; программное обеспечение; разработка программного обеспечения; Словения.

1 Introduction

Artificial intelligence (AI) has many definitions. One is "the capability of computer systems or algorithms to imitate intelligent human behavior" [1]. It can influence the process of creating new ideas and technologies that will assist in complex problem-solving and decision-making processes. Another, more technical definition states that AI is the system's ability to correctly interpret external data to learn from it and achieve predefined goals [2].

Artificial intelligence has become a significant trend in nearly all spheres of human life in recent years. Science is also following this trend. AI, its applications for various purposes, such as data analysis and quality assurance [3], and its impact are being studied in all scientific and professional areas today.

Computer science and its software engineering branch are no exceptions. The latter can be defined as "the application of a systematic, disciplined, quantifiable approach to the development, operation, and maintenance of software; that is, the application of engineering to software" [4]. This branch plays a crucial role in the digitalization of modern economics, which is of critical importance for its development and provides significant competitive advantages [5]. For example, the characteristics of software suppliers can influence the intention of small- and medium-sized enterprises (SMEs) to use modern technologies, such as data mining [6]. The software engineering and development sphere is currently undergoing significant changes driven by AI. Statistics from well-known software development companies indicate that the use of AI tools has increased significantly over the last five years. There are several reasons for this, including reduced operational costs, improved code quality, and enhanced work efficiency (Haan, 2024; Nikitashin, 2024; Worldmetrics.org., n.d). These changes also raise numerous opportunities and concerns, primarily related to the replacement of job positions, as well as the quality and security of data utilized in AI development [7]. Numerous researchers have extensively studied these aspects. Software development should be examined in both academic and professional contexts, as differences may exist between research findings and business practice [8].

In general, software development methodologies can be categorized into traditional (Waterfall), Agile (Scrum, Extreme Programming or XP, and Kanban), and Lean approaches [9]. The Lean methodology is grounded in systematic planning and emphasizes the elimination of non–value-adding activities to optimize the overall value stream. Conversely, the Waterfall model represents a sequential and plan-driven approach that prioritizes comprehensive upfront planning, wherein the input of each development phase is contingent upon the outputs of its predecessor [10]. The Agile paradigm facilitates accelerated and iterative delivery through continuous development cycles and active collaboration with end users. Numerous systematic reviews and empirical surveys [9; 11; 12] consistently identify Scrum, XP, and Kanban as the most widely adopted and extensively examined frameworks within industry practice, providing strong evidence that Agile methodologies constitute the dominant family of approaches in contemporary software development.

Shortly, the first one is an agile framework used for project management through iterative and incremental development cycles. As a form of agile project management, Scrum supports work in dynamic environments by dividing tasks into short, repeatable iterations. These provide frequent opportunities for evaluation and feedback [9; 11; 13; 14]. XP is an agile software development methodology created to support fast, high‑quality software delivery in environments with rapidly changing requirements. XP achieves its goals by applying proven engineering practices at an intensified level. Those include pair programming, test‑driven development, simple design, collective code ownership, continuous integration, and small releases. Research shows that while organizations often tailor XP to their specific needs, the method in general enhances adaptability, teamwork, and quality [13; 14]. Kanban is a framework in which the most important and well-known artifacts are representative task boards and cards. It focuses on work visualization, work in progress limitation, and flow optimization through incremental, evolutionary change [15]. Very often, it is implemented and used together with Scrum [8].

Currently, there is a lack of research in the business world that compares them with academic studies or other reliable research on the topic of artificial intelligence in the software engineering sphere. This is valid at the global level and particularly in some specific countries and regions, such as Slovenia and Eastern Europe. For example, one of the regional research projects has identified that the utilization of AI brings a large number of benefits, outweighing potential drawbacks [16]. However, problems such as a lack of knowledge and awareness of the availability of AI tools were identified as reasons for stopping their integration and utilization [16]. The above can be defined as a research gap in this work. The results of the potential research proposed at the beginning can be helpful for different purposes, like potential analysis of integration of AI, for preparation and formulation of a strategy for such integration, both in Slovenian and international software development companies, or to help solve the regionally specific problems with the familiarity and awareness described before. For this reason, we decided to conduct such applied research. The research goals are to resolve the previously described gap and prepare a base for further research on the topic at the global and local levels, as well as to assist in the formulation of recommendations and strategies in Slovenia, similar countries, and globally for the introduction of AI in software development and the resolution of related problems. The contribution of such a research is to expand existing knowledge on the topic, serving as a basis for the creation of recommendations and instructions for implementation, as well as implementation strategies themselves in Slovenia and similar countries, taking into account national and regional specificities. All of the above was poorly studied at the time of the research.

The research is sequential and consists of two parts: a review of scientific literature and reliable sources, and in-depth interviews with representatives of Slovenian companies, conducted in the form of structured interviews with software development professionals.

The first part analyzes the opinions of the scientific world and theoretical investigations on the theme. The primary purpose of the first part is to lay the groundwork for the structured interviews. Additionally, its less important goal is to try to systemize existing theoretical knowledge and previous works on the topic.

The second part examines aspects such as the popularity of AI utilization, its reasons, purposes, and impact on various aspects of the software development process, as well as the general opinion of professionals regarding AI's influence and utilization. Additionally, differences in the aforementioned aspects between different generations are examined.

Both parts have predefined research questions that guided the course of the entire research instead of hypotheses.

2 Methodology

2.1 The Background

The theme selected for the research is relatively new at the moment of writing, and some gaps related to it remain (Haan, 2024; Worldmetrics.org., n.d). One of the main reasons mentioned in the Introduction is the fact that the level of integration and utilization of AI tools in software development has grown significantly, but only in recent years (Haan, 2024; Worldmetrics.org., n.d). For this reason, it remains problematic to conduct a comprehensive bibliographic analysis or systematic literature review on the theory.

The software development sector, both globally and in Slovenia, is experiencing growth [17]. However, on the local level, this market remains relatively small. A large number of companies in the national market are SMEs. Additionally, many of them are somehow related to governmental and public institutions or have other limitations related to sharing their business data [17]. For these reasons, it is problematic to conduct comprehensive statistical research there, as the organizations and their employees are difficult to reach or lack the will or opportunity to share their data by participating in non-obligatory surveys. Additionally, companies in the Eastern European region exhibit specific trends that differ from global trends [16]. A questionnaire is usually highly standardized and has a limited number of responses and depth. A structured interview is standardized as well, but with the flexibility to ensure understanding, allowing for clearer clarification of misunderstandings, richer, more detailed responses, and better suitability for complex topics in general [18; 19]. For these reasons, the formulation of some appropriate closed questionnaires is also problematic, and more open qualitative methodologies, like the scoping literature review and the structured interview mentioned, are more suitable for the proposed research [18; 20; 19; 21].

2.2 Literature Review Methodology

As a methodology for research of the theory, the scoping literature review was used [20; 21]. The reasons were the novelty of the theme studied, as described above, and the intention to provide a comprehensive and complete overview of the topic.

The goal of the selected methodology is to identify, map, analyze, and summarize as many sources as possible on the theme of study, ensuring that no existing knowledge is overlooked and identifying key concepts and gaps [20; 21]. The most important steps here are as follows:

1. Research questions definition
2. Search query definition
3. Search for sources
4. Relevant sources selection (filtering)
5. Data extraction and organization
6. Data analysis and summarization
7. Results synthesis and presentation

The main research question was "How does artificial intelligence influence the software development sphere in Slovenia?". Considering the question and the wide utilization of agile methodologies in IT companies dominating in modern software engineering, the keywords for the search query were as follows:

• Agile methodologies
• Artificial intelligence
• Software development
• Slovenia

For the search, the keywords were combined in pairs. For each specific request, they were connected using the AND operator. Additionally, considering the rapid pace of change in the studied market, a limitation was set on the search to very recent works, no older than 5 years.

The search was conducted in the Scopus database due to its wide recognition and multidisciplinary nature. The selected database is recognized and reliable, and at the same time more comprehensive and less over-selective or subject-specific compared to alternatives (Web of Science, IEEE Xplore,...). Also, some important relevant sub-sources of the research items found were studied and included in the analysis. Considering the novelty of the theme, where it was not possible to find sufficient scientific resources, well-recognized and reliable professional sources were analyzed as a substitute for the academic ones. Professional sources, such as journals or conference proceedings, are usually reliable and peer-reviewed, but they may also cover some topics that are overlooked or lacking in academia.

It is important to note that there may be some limitations due to the selection of only one Scopus database and the lack of certain sources in it. There may also be an impact of time constraints and the resulting changes, as the literature search was prepared in the second half of 2024.

Based on the research question, the keywords, and relevance requirements, the criteria for the inclusion/exclusion were the following:

• The work is not older than 5 years at the moment of the search. More recent items were prioritized.

Goal: Ensure that only recent and up-to-date knowledge is included in the study.

• The research published in peer-reviewed resources was prioritized. Validation was based on the information from the official webpage of the resource.

Goal: Ensure the results of the analyzed work are reliable and officially confirmed.

• Keywords of the work include a combination of the query ones. Found items with a larger number of matching keywords were prioritized. At least one strict match must have been present.

Goal: Ensure the work is formally relevant. Fasten and automate the filtering of non-relevant items without the necessity to study them in depth.

• The work analysis of one of the themes declared by the research question, or its relations with the agile methodologies.

Goal: Ensure the work is relevant to the intended analysis.

Additionally, for the quality appraisal, the Critical Appraisal Skills Programme (CASP) framework, along with a standard set of questions, was used.

For the analysis and synthesis of data, quantitative and qualitative approaches are available. In our case, the quantitative approach was selected due to the area's width and relatively general criteria for the search.

This part of the research did not present any direct or explicit ethical issues or biases.

2.3 Research Questions

Based on the general research goals presented in the Introduction section and our literature research results, we have identified the following research questions:

1. What is the AI utilization rate in Slovenian software development companies?
2. In which specific areas is AI utilization the most popular?
3. What kinds of AI tools are the most popular?
4. What is the opinion of Slovenian software development professionals about AI utilization in their professional sphere?
5. Do opinions and technology acceptance of AI vary in different age groups?
6. Are there differences in the studied aspects between Slovenian and the world’s software development companies and trends?

2.4 Structured Interview Methodology

For research in the Slovenian market, we have selected a structured interview methodology. It is characterized by reliability, standardization, consistency, and efficiency, due to reduced bias and easier data analysis compared to other types of interviews. A structured interview reduces interviewer bias in the specific sense that it shifts more of the interview’s conduct into the "design" stage rather than leaving key decisions to the moment of interaction. It makes the interview situation more consistent across participants. Structure reduces the risk that the interviewer’s personal style or spontaneous choices disproportionately shape what is asked and how, thereby helping the interview better match the intended research purpose. In terms of the data analysis, the structured interview helps reduce interpretation bias in the narrow methodological sense that each piece of transcript can be more clearly linked back to a known question prompt, which supports more transparent transitions from data capture to analysis [19].

At the same time, the structured interviews require good preparation and planning. Using this methodology, the researcher must prepare a set of predefined, standardized questions that are asked in the same order each time. This set must be based on research objectives and questions. The number of interviews recommended is approximately 15 in general and at least 6 for a study of a phenomenon [18; 19].

Compared to alternatives, such as semi-structured interviews, structured interviews primarily optimize standardization and comparability across participants, which can be helpful when researcher wants uniform coverage of topics. It increases researcher control, which can reduce variability and is also more suitable for larger samples [18; 19].

Based on the results of the literature research, we have determined the objectives of the business world investigation and the questions we need to ask respondents to analyze key aspects of AI utilization (Table 1).

Table 1.

Interview Questions

Similar to the first part of the research, structured interviews presented no direct or explicit ethical issues or biases.

2.5 Structured Interview Participants

Considering the results of the literature review [22] and the fact that the market studied is specific in the terms described in the Background section, and it has a relatively small size in Slovenia [17], we have decided to conduct 15 to 20 interviews with professionals employed in different companies. Professionals were selected and invited randomly from companies of different sizes. However, their proportions by roles were defined to be as representative of a common agile software development team as possible: one product owner/business expert, one team lead/manager, and multiple developers [23; 24; 25].

It should be noted that although participants were randomly recruited from companies of varying sizes in sufficient numbers for the chosen methodology and the role relationships were designed to reflect a typical agile team, the results may still have limitations. These are related to the small sample size, which may limit the diversity of perspectives captured as well as the sensitivity to the precise distribution of companies, roles, and individual experiences included in the final sample.

During each interview, according to the selected methodology, we asked the interviewees the same questions in the same order, with one branch based on the response to the second question.

2.6 Structured Interview Analysis

Analysis of the responses to the structured interviews was conducted with the coding process standard for this methodology. First of all, initial, secondly, focused, and finally, theoretical coding.

At the first step, the descriptive approach was selected because the responses sometimes differed significantly in terms of length and depth.

Then, in the second step, two main criteria were selected for the focused coding. Those were relevance and frequency. The goals were to ensure actuality and objectivity through enhancing quantitative criterion.

Standard criteria for the theoretical coding include relevance (codes alignment with the research question and theoretical framework), purposeful thematizing (themes derivation from a clear understanding of the topic and purpose), analytical rigor (coding systematicity, transparency, and theoretical prove), saturation (coding continuity until no new insights emerge), and reflective interpretation (thoughtful coding interpretation, not just categorization) [18]. In the last step, the relevance criterion was selected as the main criterion for the theoretical step. The most important reason was that other standard criteria were not suitable for evaluating and comparing the results of the theoretical investigation, which was the main goal of the second part of the entire research. This criterion better supports logical connecting, deeper interpretations beyond the surface-level descriptions, and promotes analysis focus and efficiency as well as transparency [18; 19].

3 Results

3.1 Literature Review Results

After completing the search, 21 sources were identified and filtered according to the specified criteria. The number is relatively small due to the novelty of the topic mentioned. Additionally, a large number of sources refer to similar research works. These were the reasons for the previously mentioned filtering and inclusion of some sub-sources of the found items

The results of the analysis of the identified sources are presented below.

AI utilization has become more and more popular in software development worldwide in recent years. There are a number of reasons. AI tools help to improve different aspects of work in the area, from management and planning to development and coding. These tools are primarily used to automate and facilitate the completion of simple and repetitive tasks (Syyrilä & Kasurinen, 2024). This way, AI saves time and effort for more complex and creative work completion, such as strategic management and design [26].

The main benefits of utilizing AI tools, as pointed out by Karlovs-Karlovskis [27], are improved productivity and creativity, as well as error reduction. Interviews conducted in research by Giffari et al. (2024) also identify ease of work as one of the benefits for professionals.

However, at the same time, AI utilization raises several challenges, including ethical and legal considerations, integration difficulties, and the need for specifically qualified personnel [26; 28].

The primary disadvantages of integration are related to security vulnerabilities, inconsistent code quality [29], potential job displacement [27], lack of transparency [28], and issues with data quality [30].

Today, several research works are analyzing the state of the art in AI integration into software development. Currently, tools are available for code generation, debugging, optimization, and discovery, as well as documentation, testing, and simulation of various execution models. As previously identified, they significantly increase work productivity in their specific domains, especially for repetitive and straightforward tasks (Sauvola et al., 2024; Syyrilä & Kasurinen, 2024). Additionally, AI assistance is efficiently utilized for improving the learning process, natural language processing for code, prototyping [27], software analytics, software-related data analysis, and predicting potential problems and quality [31].

In recent years, certain branches of AI have experienced significant integration with software development. The most outstanding is digital image recognition. It was primarily used for human language recognition [32], physical object recognition and analysis [33; 34], and different kinds of medical support [35; 36; 37]. The number of specific tools in this sphere is extremely high, and numerous custom solutions are available. Additionally, there is a lack of scientific studies examining the utilization of specific tool rates. For these reasons, it is becoming problematic to identify the most popular among them. According to one of the market research studies, the leading tools in the sphere in recent years have been those from Google, for example, Google Lens [38]. Examples of other popular and well-known tools include those from Hitachi and Amazon Rekognition (Crawford, 2024).

Some authors have analyzed the impact of AI on specific aspects of software development. For example, Bahi et al. [39] and Martins [40] have researched the impact of artificial intelligence on agile development. The first authors have identified significant improvements in this process following the implementation of AI tools, based on an analysis of real-world use cases. The second has stated changes as significant but generally positive and necessary. Another study by Udoidiok et al. (2024) has examined specific AI tools, their potential, utilization, and impact on the software development process. As a result, the authors have identified artificial intelligence integration as a key trend in the future of the area and highlighted tools such as GitHub Copilot, Vercel’s v0, and Meta AI as important and powerful supportive tools. One more research by Li et al. [41] has analyzed potential AI integration and has specified key aspects impacting AI integration into the software development process. Some of these include employers paying for the utilization of learning tools and providing teaching materials and opportunities for employees.

Some researchers have attempted to predict the potential future of AI integration in software development. For example, Sauvola et al. [42] have identified four possible scenarios:

1. Minimal integration and automation of very basic, simple work.
2. Small integration and automation of some complex tasks.
3. High integration and replacement of some specific roles by AI support.
4. Full integration and automation of the software development process, with people staying as supervisors.

Saklamaeva and Pavlič (2023) have identified increased levels of collaboration and enhanced decision-making processes as potential improvements for expanding AI integration into the process. Besides the previously described aspects, Jain [43] have also pointed out potential spheres of AI utilization, such as DevOps tools management. Proposed are AI-driven protocols that adapt to potential vulnerabilities and threats in real-time, helping to optimize processes and their interoperability.

3.2 Structured Interview Results

We successfully conducted 18 interviews. The quantity and distribution of users and non-users were designed based on the methodology and principles of the study of phenomena therein. At least 15 interviews in total and at least 6 for the study of the phenomenon, in this case non-use of AI. The interviews were conducted online via video calls without recording to ensure compliance with privacy regulations.

Descriptive data of the respondents are presented in Table 2. The results are grouped by AI utilization. Identification numbers starting with 1 are assigned to those who do not utilize AI, and those starting with 2 are assigned to users of AI tools. The responses varied considerably for most aspects, except for age. These differences are important because similar roles in different organizations can vary significantly. Therefore, it was not possible to standardize and visualize the responses. As can be seen, non-users and non-technical specialists are primarily individuals belonging to older generations, while users and technical specialists are represented equally across different generations.

Table 2.

Descriptive Presentation of the Interviewees

The responses of non-users are presented in Table A.1 (Appendix A). The intention to use AI is distributed equally (fifty percent to fifty percent). The reasons for utilization/non-utilization differ, but it can be seen that "no need" was the most popular answer. It is worth noting that some respondents did not provide comprehensive answers for various reasons, including a lack of willingness or opportunity.

Results for AI users are presented in Tables A.2 and A.3 (Appendix A). In Table 5, grades are distributed from 1 (very negative) to 5 (very positive). Among AI users, coding support and information search were the most common purposes (Figure 1), the most popular tools are chatbots, like ChatGPT (Figure 2), and perceived effects were neutral to positive across all evaluated aspects (Figure 3).

Figure 1. Utilization Purposes

Figure 2. Tools Utilization

Figure 3. Utilization Impact

The general opinions of all interviewees are presented in Table A.4 (Appendix A). Qualitatively analyzing the responses using the structured interview approaches described in the Methodology section, it identified that the responses differed significantly in terms of intention, length, and depth. For this reason, it was hard to comment on them shortly, prepare a visualisations, and avoid explicit presentation.

Addressing the first research question ("What is the AI utilization rate in Slovenian software development companies?"), according to the results, it can be determined that more than half interviewees, 67%, are using AI-supportive tools (Table 2).

Answering the second research question ("In which specific areas is AI utilization the most popular?"), we can state that these tools are popular among technical specialists, especially developers. However, people employed in other roles within the sphere tend to avoid utilizing AI support (Table 2).

The response to the third research question ("What kinds of AI tools are the most popular?") is that AI users are primarily utilizing it for coding support and information search. The most popular tools are chatbots, especially ChatGPT. However, only half of Slovenian professionals and companies using AI support are utilizing integrated code-supportive tools, such as GitHub Copilot, or advanced generative and processing tools like Microsoft Office Copilot, Jasper, UiPath, and Oracle Analytics. Also, none of the respondents mentioned the usage of other tools identified in the literature (Figure 2).

Addressing the fourth research question ("What is the opinion of Slovenian software development professionals about AI utilization in their professional sphere?"), according to the opinion of the interviewed users of AI tools, they positively impact work speed, quality, and personal satisfaction, but have less of a significant impact on teamwork and communication (Figure 3). General opinion and acceptance are positive, especially among technical specialists, but some employees, primarily non-technicians, report a lack of knowledge, familiarity, opportunities, appropriate tools, and security concerns as reasons for avoiding AI integration. However, at the same time, a significant part of non-users showed a readiness and willingness to start utilizing it, also due to an understanding of the potential benefits. Additionally, some users have stated their own utilization issues and concerns, like imperfections, errors, and potential professional degradation.

Responding to the fifth research question ("Do opinions and technology acceptance of AI vary in different age groups?"), we can determine that, according to the results of our qualitative analysis, AI utilization doesn’t differ among different ages. Both young and old technical specialists tend to accept and utilize these tools ("European Union Age Structure - Demographics." n.d.). At the same time, non-technical employees of all generations have a more negative attitude and greater concerns. Additionally, the results of the analysis are not absolutely clear.

Concluding the previously described, we can conclude that acceptance of AI in Slovenian software development companies is primarily related to the domain and role of the employee, and only then to the age. Analyzing the opinions, we can primarily see an optimistic attitude and a willingness to accept this technology. The interviewees believe that AI already does and will continue to positively impact the sphere of software development in various aspects, such as productivity, efficiency, and ease of work. However, at the same time, AI tools still require development, optimization, and external advisory support for their results. Additionally, it is crucial to avoid overreliance on AI support and consider important concerns, such as information security.

4 Discussion

Both parts of the entire research, the literature review and the structured interviews, have their own limitations.

The scoping literature review was prepared based on a search in only one database despite its size, recognition, and multidisciplinarity. Additionally, because the AI area is evolving quickly, some differences and changes from the planned methodology, as explained in the Methodology section, have occurred. These were due to the newness of the topic and the limited number of scientific sources available at the time of the search. At the same time, this part was primarily conducted in accordance with all the most important recommendations [20; 21] and the well-recognized methodology planned, so it can be stated that it has reached its goal.

The structured interview part has included different types of companies and information solutions without consideration of industries and specializations for the reasons described in the Methodology section. This can also introduce some limitations in terms of representativeness due to potential differences between specific subareas of software engineering. However, this part was conducted in almost total accordance with the well-recognized methodology that was selected [18; 19]. It has included different kinds of professionals in realistic proportions from organizations of various sizes, so we can say that its goal was also achieved.

According to the results of our literature review, the theme of AI in software development is very popular in the scientific world at present. However, the theme is relatively new and there is a small number of theoretical works studying the business world’s situation. There are no academic works studying AI utilization in the software development sphere in some specific country, at least in the region of Central-Eastern Europe, to which Slovenia belongs both geographically and culturally, and which it represents in a typical fashion [17]. Additionally, existing scientific research works primarily focus on different aspects than those studied in our work. For these reasons, it is problematic to compare our research’s results with others in some respects.

At the same time, there are some points of intersection. Comparing interview responses with the results of the literature investigation, we can see similarities.

The AI utilization rate among our interviewees from the Slovenian software development sector is approximately the same as the world’s 62% [44], but lower than in some other countries, such as the United States, where it is 92% [45].

Some of the potential problems and disadvantages of AI utilization mentioned in the literature have also been reported by specialists. For instance, information security issues and potential problems with the quality of data and outputs [28; 29; 30]. Problems that hinder AI tool integration are similar in scientific literature and the structured interview part of our research as well. For instance, a lack of knowledge, familiarity, or suitable solutions for businesses, as well as issues with integrating into existing products [26; 28]. The opinions primarily support the results of scientific works.

The advantages are aligned as well. For instance, improved productivity of workers, quality of work, as well as its speed, ease, and employee satisfaction (Bhandari et al., 2023; Giffari et al., 2024; Karlovs-Karlovskis, 2024; Syyrilä & Kasurinen, 2024). The only exceptions are teamwork and communication. The scientific literature promises significant improvements, whilst the prevailing part of Slovenian software developers has reported no significant impact [46].

At the same time, some of the advantages or problems pointed out in the literature were not mentioned by the respondents. In parallel, some of the shortcomings, such as poor local language optimization, were not mentioned in scientific sources, but they were reported by respondents. Additionally, there are some similar inconsistencies in terms of tool utilization. Many of the specific tools, such as solutions for image recognition (Google Lens, Rekognition) or their alternatives, pointed out in the literature, were not mentioned by Slovenian professionals at all. The same applies to the solutions for management support [30]. Inconsistencies may be related to the limitations of research literature or structured interviews. However, other factors, such as the impact of country-specific conditions or misalignments between theoretical studies and the actual market situation, may also be at play. For this reason, we can formulate the response to research question 6 ("Are there differences in the studied aspects between Slovenian and the world’s software development companies and trends?") as there can be differences in AI integration and utilization in the software development sphere in Slovenia and around the globe. Here, we can consider additional, more comprehensive research to support or refute our findings and assumptions.

In summary, existing research generally analyzes and describes the market situation in software development and its relationship with AI correctly. However, some differences were identified while analyzing the Slovenian local market. These can have different reasons, including national or even regional specifics, considering that, as mentioned before, the country is typically representative of its region in Central-Eastern Europe. The reasons mentioned may be the large number of small and medium-sized companies, which have significantly fewer needs and opportunities for adapting newer technologies within them, the large public sector, which accepts and introduces changes significantly slower than the private sector and generally has more restrictions, including those in adapting the aforementioned new technologies, or the negative impact of digital piracy on the engineering and software development market in Slovenia, which reduces the economic profitability and activity of companies and consequently the possibilities for introducing technologies such as AI. This aspect had never been analyzed or identified before, at least not based on the results of our literature review. Additionally, the reasons can be different and valid not just locally, but also globally, and highlight shortcomings of previous research works in general. Here, again, we recommend additional investigations at all three levels: national, regional, and global.

The results of our research are not only useful for the development of theory but also for practical applications. For instance, to analyze the potential integration of AI into software development across all three levels mentioned previously, and to prepare and formulate a strategy for integration, considering both general principles and national and regional specifics of the modern software development market.

5 Conclusions

We can conclude that the theme of AI in software development is relatively new, yet very popular in recent years. Evaluating the results, we have studied only part of all existing scientific and professional sources. Also, the number of interviews was relatively small.

However, the sources studied are primarily very new and not outdated. The sample selected for the structured interview research is representative and of adequate size, considering the literature recommendations and the specifics of the market studied. The results of the correlational analysis are ambiguous. At the same time, they highlight the need for further investigations into the subject and serve as a good starting point for such studies.

For the reasons described above, we can confidently declare our research a success. Its results are useful in different terms. For instance, potential AI integration analysis, preparation, and strategy formulation for this integration, considering national specifics in Slovenian software development companies, or in other parts of the world. Additionally, the results can serve as a starting point for further research on the theme in Slovenia and other countries.

References:

1. Merriam-Webster. (n.d.). Artificial intelligence. In Merriam-Webster.com dictionary. Date of access: August 8, 2025, URL: https://www.merriam-webster.com/dictionary/artificial%20intelligence
2. Ambrožič, V., Krajnc, A., & Štok, B. (2024). Primer razvoja pametnega asistenta. In OTS 2024 Sodobne informacijske tehnologije in storitve. DOI: 10.18690/um.feri.4.2024.2
3. Gao, J., Agarwal, R., & Garsole, P. (2025). AI Testing for Intelligent Chatbots—A Case Study. Software, 4(2), 12. DOI: 10.3390/software4020012
4. Cho, S. S., Caporusso, N., & Doyle, M. (2024). P4: Principles, Patterns, Practices, and Projects for Effective Software Engineering Education. In 2024 47th MIPRO ICT and Electronics Convention (MIPRO) (P. 1289–1294. DOI: 10.1109/mipro60963.2024.10569934
5. Bittner, B., Kovács, T., & Nábrádi, A. (2022). The Integration of Digital Technologies in the European Union based on the DESI Index. Proceedings of the ENTRENOVA - Enterprise Research Innovation Conference, 8(1), 337–349. DOI: 10.54820/entrenova-2022-0029
6. Bach, M. P., Topalović, A., & Turulja, L. (2022). Data mining usage in Italian SMEs: an integrated SEM-ANN approach. Central European Journal of Operations Research, 31(3), 941–973. DOI: 10.1007/s10100-022-00829-x
7. King, C. (2023). A.I.’s Impact In Its First Year In Software Engineering. Forbes. Date of access: July 15, 2024, URL: https://www.forbes.com/sites/forbes-engineering/2023/12/07/ais-impact-in-its-first-year-in-software-engineering/
8. Nikitashin, M., Kaluža, M., & Werber, B. (2024). Analysis of Methodologies and Tools for Software Development in Different Architectures. In 2024 47th MIPRO ICT and Electronics Convention (MIPRO) (P. 1999–2007. DOI: 10.1109/mipro60963.2024.10569545
9. Anifa, M., Ramakrishnan, S., Kabiraj, S., & Joghee, S. (2024). Systematic review of literature on Agile approach. NMIMS Management Review, 32(2), 84–105. DOI: 10.1177/09711023241272294
10. Royce, W. (1987). MANAGING THE DEVELOPMENT OF LARGE SOFTWARE SYSTEMS: CONCEPTS AND TECHNIQUES. In ICSE ’87: Proceedings of the 9th International Conference on Software Engineering (P. 328–338). https://dl.acm.org/doi/10.5555/41765.41801
11. Al-Zewairi, M., Biltawi, M., Etaiwi, W., & Shaout, A. (2017). Agile Software Development Methodologies: Survey of Surveys. Journal of Computer and Communications, 05(05), 74–97. DOI: 10.4236/jcc.2017.55007
12. Behrens, A., Ofori, M., Noteboom, C., & Bishop, D. (2021). A systematic literature review: how agile is agile project management? Issues in Information Systems, 22(3. DOI: 10.48009/3iis2021298-316
13. Rahul, N. A., Nouidui, T. S., Ulaya, P. L., & Kiwia, D. D. (2023). The Impact of Agile Methods on the Software Projects Implementation and Management. American Journal of Industrial and Business Management, 13(04), 183–194. DOI: 10.4236/ajibm.2023.134013
14. Sassa, A. C., Alves De Almeida, I., Fernandes Pereira, T. N., & Silva De Oliveira, M. (2023). Scrum: A Systematic Literature Review. (IJACSA) International Journal of Advanced Computer Science and Applications, 14(4). https://thesai.org/Downloads/Volume14No4/Paper20-ScrumASystematicLiteratureReview.pdf
15. Al-Baik, O., & Miller, J. (2014). The kanban approach, between agility and leanness: a systematic review. Empirical Software Engineering, 20(6), 1861–1897. DOI: 10.1007/s10664-014-9340-x
16. Nikitashin, M. (2025a). Artificial Intelligence and Management of Dualities in Software Development Companies. Uporabna informatika, 33(1), 19-28. DOI: 10.31449/upinf.247
17. Nikitashin, M. (2025b). Economics of Software Engineering in Slovenia. Revija za ekonomske in poslovne vede, 12(1), 18–38. DOI: 10.55707/eb.v12i1.148
18. Bartholomew, T. T., Joy, E. E., Kang, E., & Brown, J. (2021). A choir or cacophony? Sample sizes and quality of conveying participants’ voices in phenomenological research. Methodological Innovations, 14(2. DOI: 10.1177/20597991211040063
19. Kvale, S. (1996). InterViews: An introduction to qualitative research interviewing. Sage Publications.
20. Hart, C. (1998). Doing a Literature Review: Releasing the Social Science Research Imagination. In Sage eBooks. http://ci.nii.ac.jp/ncid/BB14383612
21. Ma, Z. G. (2022). Scoping Review in One Week: A cookbook for beginners to conduct scoping reviews and write review papers with simple steps. https://portal.findresearcher.sdu.dk/en/publications/scoping-review-in-one-week-a-cookbook-for-beginners-to-conduct-sc/
22. Elhami, A., & Khoshnevisan, B. (2022). Conducting an Interview in Qualitative Research: The Modus Operandi. Mextesol Journal., 46(1), 1–7. DOI: 10.61871/mj.v46n1-3
23. Schwaber, K. (2004). Agile project management with Scrum (Developer Best Practices). Wash. Microsoft Press.
24. Siau, K., Tan, X., & Sheng, H. (2007). Important characteristics of software development team members: an empirical investigation using Repertory Grid. Information Systems Journal, 20(6), 563–580. DOI: 10.1111/j.1365-2575.2007.00254.x
25. Sæter, G. E., Stray, V., Almås, S., & Lindsjørn, Y. (2024). The Role of Team Composition in Agile Software Development Education: A Gendered Perspective. In Lecture Notes in Business Information Processing (P. 179–195. DOI: 10.1007/978-3-031-61154-412
26. Bhandari, K., Kumar, K., & Sangal, A. L. (2023). Artificial Intelligence in Software Engineering: Perspectives and Challenges. In 2023 Third International Conference on Secure Cyber Computing and Communication (ICSCCC) (P. 133–137. DOI: 10.1109/icsccc58608.2023.10176436
27. Karlovs-Karlovskis, U. (2024). Generative Artificial Intelligence Use in Optimising Software Engineering Process: A Systematic Literature Review. Applied Computer Systems, 29(1), 68–77. DOI: 10.2478/acss-2024-0009
28. Chergarova, V., Tomeo, M., Morgan, H., & Andrade, S. (2024). Integrating artificial intelligence (AI) into game development to elevate diverse gameplay elements. Issues in Information Systems, 25(2), 408–417. DOI: 10.48009/2iis2024132
29. Coutinho, M., Marques, L., Santos, A., Dahia, M., França, C., & De Souza Santos, R. (2024). The Role of Generative AI in Software Development Productivity: A Pilot Case Study. In AIware ’24: 1st ACM International Conference on AI-Powered Software. DOI: 10.1145/3664646.3664773
30. Udoidiok, I., Reza, H., & Zhang, J. (2024). Exploring AI Integration in Software Development: Case Studies and Insights. In NAECON 2024 - IEEE National Aerospace and Electronics Conference. DOI: 10.1109/naecon61878.2024.10670347
31. Caldeira, J., Abreu, F. B. E., Cardoso, J., Simões, R., Oliveira, T., & Reis, J. P. D. (2023). Software Development Analytics in Practice: A Systematic Literature Review. Archives of Computational Methods in Engineering, 30(3), 2041–2080. DOI: 10.1007/s11831-022-09864-y
32. Mukhanov, С., Rahim, A. A., Yemberdiyeva, A., & Bazarbekov, I. (2024). KAZAKH SIGN LANGUAGE RECOGNITION BY USING MACHINE LEARNING METHODS. Vestnik Almatinskogo universiteta energetiki i svyazi, 2(65), 131–148. DOI: 10.51775/2790-08862024652131
33. Liu, M., Zhang, X., Wang, X., Wei, Z., & Cheng, X. (2024). RESEARCH ON MAIZE STEM RECOGNITION BASED ON MACHINE VISION. Engenharia Agrícola, 44. DOI: 10.1590/1809-4430-eng.agric.v44e20240028/2024
34. Moussaoui, H., Akkad, N. E., Benslimane, M., El-Shafai, W., Baihan, A., Hewage, C., & Rathore, R. S. (2024). Enhancing automated vehicle identification by integrating YOLO v8 and OCR techniques for high-precision license plate detection and recognition. Scientific Reports, 14(1. DOI: 10.1038/s41598-024-65272-1
35. Nikishina, V., Petrash, E., Engel, E., Simonenko, I., & Shagina, E. (2024). Spatial Orientation Parameters of Face Image Recognition as Predictors of Liver Failure Symptoms in Adolescents. Bulletin of Russian State Medical University, 2024(2. DOI: 10.24075/brsmu.2024.015
36. Yoda, H., Abe, K., Takeo, H., Takamura-Enya, T., & Koike-Takeshita, A. (2024). Application of image-recognition techniques to automated micronucleus detection in the in vitro micronucleus assay. Genes and Environment, 46(1. DOI: 10.1186/s41021-024-00305-9
37. Zhang, K. C., Qiao, Z., Yang, L., Zhang, T., Liu, F. L., Sun, D. C., Xie, T. Y., Guo, L., & Lu, C. R. (2024). Computer-vision-based artificial intelligence for detection and recognition of instruments and organs during radical laparoscopic gastrectomy for gastric cancer: a multicenter study. PubMed, 27(5), 464–470. DOI: 10.3760/cma.j.cn441530-20240125-00041
38. Grand View Research. (2022). Image recognition market size, share & trends analysis report by technique, by component (hardware, software, service), by deployment mode, by vertical, by application (augmented reality, security & surveillance), by region, and segment forecasts, 2024-2030 [Industry report]. Grand View Research. Date of access: August 15, 2025, URL: https://www.grandviewresearch.com/industry-analysis/image-recognition-market
39. Bahi, A., Gharib, J., & Gahi, Y. (2024). Integrating Generative AI for Advancing Agile Software Development and Mitigating Project Management Challenges. International Journal of Advanced Computer Science and Applications, 15(3. DOI: 10.14569/ijacsa.2024.0150306
40. Martins, M. R. (2023). From Traditional IT Implementations to AI-Driven Projects. International Journal of Scientific Research and Management (IJSRM), 11(07), 5011–5028. DOI: 10.18535/ijsrm/v11i07.em03
41. Li, Z. S., Arony, N. N., Awon, A. M., Damian, D., & Xu, B. (2024). AI Tool Use and Adoption in Software Development by Individuals and Organizations: A Grounded Theory Study. arXiv (Cornell University. DOI: 10.48550/arxiv.2406.17325
42. Sauvola, J., Tarkoma, S., Klemettinen, M., Riekki, J., & Doermann, D. (2024). Future of software development with generative AI. Automated Software Engineering, 31(1. DOI: 10.1007/s10515-024-00426-z
43. Jain, S. (2023). Integrating Artificial Intelligence with DevOps: Enhancing continuous delivery, automation, and predictive analytics for high-performance software engineering. World Journal of Advanced Research and Reviews, 17(3), 1025–1043. DOI: 10.30574/wjarr.2023.17.3.0087
44. Stack Overflow. (2024). AI [Section of the 2024 Stack Overflow Developer Survey]. Stack Overflow Developer Survey. Date of access: August 17, 2024, URL: https://survey.stackoverflow.co/2024/ai#1-ai-tools-in-the-development-process
45. Shani, I. (2024). Survey reveals AI’s impact on the developer experience - The GitHub Blog. The GitHub Blog. Date of access: August 17, 2024, URL: https://github.blog/news-insights/research/survey-reveals-ais-impact-on-the-developer-experience/
46. Saklamaeva, V., & Pavlič, L. (2023). The Potential of AI-Driven Assistants in Scaled Agile Software Development. Applied Sciences, 14(1), 319. DOI: 10.3390/app14010319
47. Crawford, J. (2024, June 24). Best Image Recognition Tools for Your Business. Tech Research Online. Date of access: August 15, 2024, URL: https://techresearchonline.com/blog/10-best-image-recognition-tools-for-2022/
48. European Union Age structure - Demographics. (n.d.). Date of access: August 15, 2024, URL: https://www.indexmundi.com/europeanunion/agestructure.html
49. Giffari, R., Ridho, M. M., Sensuse, D. I., Hidayat, D. S., & Purwaningsih, E. H. (2024a). Analyst’s Perception on the Use of AI-based Tools in the Software Development Life Cycle. Jurnal Sistem Informasi, 20(1. DOI: 10.21609/jsi.v20i1.1399
50. Haan, K. (2024). 22 Top AI Statistics And Trends. Forbes Advisor. Date of access: July 15, 2024, URL: https://www.forbes.com/advisor/business/ai-statistics
51. Nikitashin, M. (2024). AI and UI Development: A Case Study of Mobile Application X. Uporabna informatika, 32(4), 180-186. DOI: 10.31449/upinf.244
52. Syyrilä, E., & Kasurinen, J. (2024, May 20). Developing Smarter Software Engineering Tools by Utilizing AI Assistance. In 2024 47th MIPRO ICT and Electronics Convention (MIPRO. DOI: 10.1109/mipro60963.2024.10569365
53. Worldmetrics.org. (n.d.). Market Research, Statistics & Business Insights • GitNux. Date of access: July 15, 2024, URL: https://worldmetrics.org/ai-in-software-development-statistics/

Appendix A

Table А.1.

Responses of the Non-users

Table А.2

Responses of the Users to the Open Questions

Table А.3

Responses of the Users to the Rating Questions

Table А.4

General Opinions of the Interviewees