Integrating AI Tools Into Learning: A Study of Students’ Attitudes in Universities, Colleges, and Vocational Institutions

2.1 Use of AI tools among students

AI tools are becoming increasingly prevalent in student learning due to their accessibility and ability to answer questions quickly. Research shows that students actively use AI tools in both vocational and higher education, with ChatGPT emerging as the most popular platform (Annuš, 2025; Gesser-Edelsburg et al., 2024; Jereb & Urh, 2024; Metreveli et al., 2025). A Digital Education Council (DEC) survey of 3,839 undergraduate, Master’s and PhD students from 16 countries found that 86% of respondents use AI during their studies: 24% do so daily and 54% at least once a week, indicating a high level of engagement with these technologies (DEC, 2024). Similar patterns were observed in Slovenia, where 24% of students use AI several times a week and 9% almost every day (Jereb & Urh, 2024). In a USA study conducted at a university offering doctoral programs, all 68 surveyed students reported being familiar with ChatGPT (Divekar et al., 2024). In Germany, a survey of 108 students found that 96.4% were aware of ChatGPT, 40.2% used AI more than twice per week for academic purposes, and 17% used it almost daily (Kubullek, Kumaç & Dogangün, 2024). In a vocational and technical college in mainland China (N = 756), 61.4% of respondents reported having prior experience using generative AI (GenAI) tools for learning (Wu, Li & Mo, 2025).

Among the most frequently used tools, ChatGPT (66%), Grammarly and Microsoft Copilot (25% each) stand out, with more than two AI tools used per student on average (DEC, 2024). In addition to these applications, students also use Google Gemini, Claude AI, Perplexity, DeepL and visual content creation tools such as Canva with its AI features DALL-E or Midjourney.

Higher Education. International studies demonstrate a consistent trend in the use of AI in higher education. A DEC study (2024) found that information retrieval is the most popular use case of AI among students, followed by grammar checking, summarising documents, paraphrasing texts, and creating first drafts. A study conducted in Slovenia (N = 258) showed that students most frequently use AI for information retrieval and idea generation, and as many as 86.3% of respondents reported that AI helps them to better understand the content of their studies (Jereb & Urh, 2024). A Hong Kong study (N = 399) revealed positive attitudes towards GenAI in teaching and learning processes, recognising its potential for personalised learning, writing and idea generation support and research analysis (Chan & Hu, 2023). AI tools, particularly LLM models, enable students to review and process large amounts of information more quickly, which speeds up literature reviews and the initial stages of data analysis, allowing more time for deeper thinking and critical evaluation (Chan & Hu, 2023). A study at Bentley University in the USA (N = 68) revealed the active use of AI in academic work and in learning about complex topics (Divekar et al., 2024). A qualitative study in Finland (N = 12) showed that students value ChatGPT as a useful tool for understanding academic concepts and as an additional learning resource that allows them to get quick explanations and help with coursework without waiting for the lecturer’s answers (Fuchs & Aguilos, 2023). A study conducted by Malaysian researchers (N = 284) confirmed that AI increases student engagement, helps to solve academic problems and keep students up to date (Khairuddin et al., 2024).

Vocational Education and Training. The integration of AI into vocational education and training (VET) systems remains less explored compared to higher education, despite its growing importance for future labour market demands (Annuš, 2025; Wu, Li & Mo, 2025). Research from Malaysia highlights the potential of virtual simulations in vocational training, demonstrating their effectiveness in providing practical, risk-free experience and improving students’ competitiveness in real-world professional settings (Amdan et al., 2024).

A study conducted in Slovakia (N = 200) revealed that IT-focused vocational school students primarily used generative AI for debugging and code generation. Interestingly, both beginner and advanced learners reported similar levels of perceived usefulness, suggesting that AI tools are viewed as broadly supportive across programming tasks, regardless of experience level (Annuš, 2025). In Sweden, a case study (N = 14) showed that students used AI tools to learn new skills, find and analyse information, generate ideas, and support professional tasks such as programming, text editing, and marketing. The programme also helped students better understand how AI works, recognise its business value, and explore new career opportunities (Metreveli et al., 2025).

In Ukraine, research (N = 200) revealed that AI adoption in vocational education is still at an early stage. Most integration occurs through distance learning platforms such as Moodle, which do not yet fully use AI technologies (Vasiutynska et al., 2025). The study also found major gaps compared with global levels, especially in funding (25% local vs. 85% global) and AI integration (60% vs. 85%), showing a clear shortage of resources for innovation (Vasiutynska et al., 2025).

Globally, AI integration in vocational education focuses on personalised learning, the use of generative assistants and virtual tutors, real-time career guidance, and the automation of training and administrative tasks (Amdan et al., 2024; Duan & Wu, 2024; Ejjami, 2024; Jiang et al., 2024; Vasiutynska et al., 2025; Wu, Li & Mo, 2025). AI also supports the use of simulations and virtual reality, giving students practical, risk-free experience and improving their employability (Ejjami, 2024).

Across different contexts, AI use in vocational education is linked to positive learning outcomes, higher motivation and engagement, and better individual learning support through adaptive feedback and context-based content generation (Amdan et al., 2024; Çela et al., 2024; Ejjami, 2024; Suparyati et al., 2023; Treve, 2024).

2.2 Challenges for the integration of AI in studies

AI technologies, especially GenAI, such as ChatGPT, have opened a new era in the academic sector (Bozkurt, 2024a; Gesser-Edelsburg et al., 2024). Although AI applications such as intelligent learning systems and learning analytics have been used in education for some time, the emergence of GenAI in 2022 marked a turning point – one characterised by both optimism and deep concern (Bozkurt, 2024a, 2024b; Bozkurt et al., 2021). It is therefore essential that GenAI integration ensures that these tools complement the learning process, rather than undermine critical thinking and academic integrity (Bae & Bozkurt, 2024; Gesser-Edelsburg et al., 2024).

The main challenges related to the implementation of AI into education are ethical risks, the need for AI literacy and critical thinking, and the lack of clear institutional guidelines.

Ethical considerations and risks related to the quality of learning. Despite the benefits of AI, the academic community faces challenges posed by AI. UNESCO emphasises that students should understand AI ethics, including responsible use and data privacy. A survey by the DEC (2024) shows that 50% of students are concerned about how the use of AI tools may affect academic integrity, 60% question the fairness of AI assessment, 61% are concerned about privacy and data security, and 51% are concerned about the reliability of AI-generated content (DEC, 2024). Students also express concerns about bias in AI decisions (32%), insufficient AI competences of lecturers (28%) and a possible decline in the quality of education (27%). In addition, 25% of students have concerns about accessibility and inclusion, and believe that AI may reduce the value of their degree, while 23% question the return on their educational investment. Over-reliance on AI is also a concern: 55% of students believe it reduces the quality of teaching and 52% say it negatively affects their academic performance (DEC, 2024). Jereb and Urh (2024) note that students face problems with the accuracy and fact-checking of AI, and the tools are not always effective in dealing with complex tasks. Furthermore, a Finnish study (Fuchs & Aguilos, 2023) highlights the risk of plagiarism, with students easily copying AI-generated content, and some admitting to using ChatGPT in defiance of institutional rules. The easy availability of quality AI responses encourages academic dishonesty, which challenges the quality of learning. Chen and Lin (2024) point out that access to a wide range of learning resources through AI stimulates students’ creativity and allows for deeper exploration of topics but can weaken critical thinking if used irresponsibly. Khairuddin et al. (2024) stress that excessive use of AI can negatively affect creativity and independent thinking. Researchers suggest that AI should be seen as a tool for generating ideas and exploring different perspectives, emphasising the importance of developing independent thinking and the ethical aspects of using AI (Lee & Low, 2024; Lubbe, Marais & Kruger, 2025). While AI provides useful tools that can support certain stages of research, the final insights, analysis and conclusions should be based on students’ own critical thinking and academic integrity (Slimi, 2023). AI tools, such as ChatGPT, should be used as complementary tools rather than as the main source to ensure the quality and academic rigour of research texts (Bozkurt, 2024a; Fuchs & Aguilos, 2023; Gluoksnyte, 2025). Despite the risks, many students do acknowledge AI’s benefits, with 84.3% believing it enhances the quality of their studies (Jereb & Urh, 2024). However, traditional learning methods continue to be highly valued. In a UK study (N = 302), students still value traditional teaching more highly on several academic dimensions: academic support, assessment, encouragement to perform better and course quality (Rodway & Schepman, 2023).

AI literacy and critical thinking in education. Effective integration of AI in education requires a high level of AI literacy and critical thinking. AI literacy is defined as the ability to understand and effectively use AI tools while developing resilience to the challenges posed by AI-based environments (Zhao, 2025). It is important that AI does not become a substitute for students’ thinking; on the contrary, students should improve their critical thinking and evaluation skills by constantly evaluating the content created by AI. This helps AI become an aid rather than a substitute for thinking (Nagelhout, 2023). Higher-level thinking skills, such as analysis, evaluation, and creation (according to Bloom’s taxonomy), are becoming increasingly important in an AI-driven world (Lubbe, Marais & Kruger, 2025).

Contemporary research reveals a significant knowledge gap regarding AI and GenAI within the academic community (Gesser-Edelsburg et al., 2024). For example, DEC (2024) found that 58% of students do not believe they have sufficient knowledge about AI, and 48% do not believe they are adequately prepared for a job market where AI is widely used. In addition, 80% say that the use of AI in universities does not meet their expectations. Students emphasise the need for training: 72% would like training on the effective use of AI tools, 73% believe that such training is necessary for teachers, and 72% would like more courses on AI literacy (DEC, 2024). Divekar et al. (2024) emphasise that AI literacy is essential to ensure the responsible and effective use of technology. Research also shows that insufficient knowledge of AI tools is one of the main barriers to their effective use (Gesser-Edelsburg et al., 2024).

Student involvement in AI integration processes is particularly important. According to DEC (2024), 71% of students want to participate in decision-making regarding AI implementation in universities, which indicates a desire for greater involvement. Students expect universities to organise training for teachers so that they can more effectively integrate AI tools into the teaching process (DEC, 2024). It is useful for graduates to know how to effectively integrate GenAI into their work. This means being able to apply expert knowledge and professional judgment rather than blindly accepting AI results. Engaged augmentation is essential – the strategic and conscious use of AI technologies in education to enhance human capabilities, and promote critical thinking and creativity, rather than passively adapting or allowing technologies to dominate (Dahri et al., 2024; Gesser-Edelsburg et al., 2024). This requires both students and educators to be AI literate and actively involved in AI integration processes, ensuring that technologies align with educational values and contribute to the creation of engaging and effective learning environments (Farooqi, Amanat & Awan, 2024; Gesser-Edelsburg et al., 2024).

Guidelines and communication on the use of AI. The lack of clear guidelines for the use of AI still remains a significant challenge. According to DEC (2024), only 5% of students are fully aware of their university’s AI guidelines and consider them to be comprehensive, while 86% are not well informed and 93% believe that these guidelines should be improved. These results highlight communication gaps and the need for clearer rules. Divekar et al. (2024) argue that institutions must develop clear guidelines that maximise the potential of AI and minimise risks. Khairuddin et al. (2024) suggest integrating AI into teaching strategies to ensure responsible use, while Fuchs and Aguilos (2023) emphasise that universities should introduce plagiarism prevention measures, train in the responsible use of AI, and adapt teaching methods to encourage critical thinking and deeper engagement. Zhang and Aslan (2021) highlight the gap between the potential of AI technologies and their actual application in education. They call for more empirical research in real learning environments to better address educational goals and needs. The challenge for institutions is not only to provide training, but also to develop systems to prevent dependence on AI and ethical problems.

While the scientific literature confirms the significant opportunities AI tools offer for learning and increased engagement, it also highlighted the significant lack of AI literacy and challenges related to institutional readiness. Most studies have focused on higher education institutions, with less attention paid to vocational education, where digital readiness and pedagogical contexts differ significantly (Ejjami, 2024; Wu, Li & Mo, 2025).

Building on the findings of previous studies, the aim of this study is to provide empirical evidence from the Lithuanian context, examining institutional and disciplinary differences in the implementation of AI, its perceived impacts, and ethical aspects. The survey design directly reflects the insights of the literature review – examining topics such as AI literacy, critical thinking, academic integrity, and institutional readiness – with the aim of broadening the discussion beyond higher education and providing new comparative data for the international field.

4.1 Using AI tools

Frequency of AI tool use. (1–7 scale: ‘Never’ → ‘Daily’). Overall, ChatGPT is the most commonly used tool (M = 4.15), corresponding to usage ranging from ‘Monthly’ to ‘Several times a week’. In contrast, all other tools are used infrequently, including DeepL (M = 1.87), Grammarly (M = 1.58), Gemini (M = 1.57), GitHub Copilot (M = 1.38), and Image Creator/Bing (M = 1.36).

By institution, universities (M = 4.31, SD = 1.72) and colleges (M = 4.42, SD = 2.03) report similarly high ChatGPT use, though colleges show greater variability across students (higher SD). Vocational schools use ChatGPT markedly less (M = 3.45, SD = 2.03), typically between ‘Once per semester’ and ‘Once per month’. ChatGPT dominates students’ AI usage across sectors, with lower uptake in vocational schools and greater heterogeneity in colleges. Other AI tools are currently niche in study-related use.

Use of AI tools in different types of educational institutions. Of the 803 students surveyed, 570 (71%) reported using AI tools in their studies, including ChatGPT, code assistants, and AI-based learning platforms: 163 college students (75.5% of college students), 327 university students (78.8% of university students) and 80 vocational school students (46.5% of vocational school students). Of the surveyed students, 199 (24.8%) do not use AI tools in studies and 34 (4.2%) are not aware of the possibility of doing so. It is also noticeable that ignorance of AI tools is highest among vocational students, 24 out of 172 (14%), compared to students from colleges (1 out of 216, 0.5%) and universities (9 out of 415, 2.2%) (Figure 2).

Figure 2

The results show that while university students are the most frequent users of AI tools, college students are also active users, while vocational students are the least frequent users. This is confirmed by the statistically significant difference between the users of AI tools in different educational institutions (Fisher’s Exact Test returned p = 2.2e–16). The type of educational institution influences the use of AI tools, but the differences are not very pronounced (Cramér’s V = 0.24). This means that the relationship between the educational institution and the use of AI tools is moderate. Thus, the type of training institution has a statistically significant effect on the use of AI tools. These results show that students from higher education institutions are more likely to use AI tools than students from vocational schools, but the trend is similar between university and college students.

The use of AI tools in higher education institutions among students from different years of study. A chi-square test showed a statistically significant relationship between the year of study and the use of AI tools (Chi², p = 0.012), suggesting that AI tool usage varies by study year. The association, although statistically significant, was of low strength (Cramér’s V = 0.12).

The use of AI tools in higher education institutions among students from different fields of study. The use of AI tools is unevenly distributed across the different fields of study, with the highest levels of use recorded in computer science (90.3%), arts (83.3%), health sciences (82.1%) and social sciences (80.7%). The lowest rate of AI use was observed in educational sciences at 66.7%, although here too the majority of students reported using these tools. Statistical analysis of the results revealed a statistically significant relationship between the field of study and the use of AI tools (Chi², p = 0.001, Cramer’s V = 0.17). The percentage of students that do not know if they use AI tools is relatively low in all fields of study, except for arts, where the figure is 16.7%.

4.2 Students’ attitudes towards the impact of AI tools on learning

Students’ perceptions of the impact of AI tools on learning in different types of educational institutions. Students’ attitudes towards the impact of AI tools such as ChatGPT on learning were compared in three different educational settings: colleges, universities and vocational schools. Most respondents (43%, N = 345) believe that there is a balance between positive and negative impacts. This opinion is prevalent in all educational institutions, but slightly more pronounced in universities (46%, N = 191) and colleges (44.4%, N = 96), and less pronounced in vocational schools (33.7%, N = 58). The positive impact of AI is most perceived by almost a third (27.5%, N = 221) of students in all types of educational institutions. This view is most common in colleges (31.5%, N = 68), less common in universities (26.7%, N = 111) and least common in vocational schools (24.4%, N = 42). A small proportion of respondents (13.2%, N = 106) consider the impact of AI to be more negative. This attitude is most pronounced in universities (15.4%, N = 64), to a lesser extent in vocational schools (13.4%, N = 23) and to a lesser extent in colleges (8.8%, N = 19). A significant proportion of respondents (16.3%, N = 131) indicated that they did not have a clear opinion, indicating that some of them are not yet sufficiently familiar with AI tools or do not feel that they have a significant impact on their studies. Interestingly, this opinion is most prevalent in vocational schools (28.5%, N = 49) and least prevalent in universities (11.8%, N = 49) (Figure 3).

Figure 3

These findings show that students’ attitudes towards the impact of AI on higher education depend on their institution of study. University and college students are more likely to see a balanced or positive impact, while vocational students are more likely to express doubts or see more negative aspects. Statistical analysis confirms that these differences are statistically significant (Chi², p < 0.001, Cramer’s V = 0.14), implying that students from different institutions have statistically significant differences in their perceptions of the impact of AI models on learning.

Higher education students’ perceptions of the impact of AI tools on their learning by field of study. The study assessed university and college students’ views on the impact of AI tools on learning. Figure 4 shows the distribution of responses (N = 631) from university and college students regarding the impact of AI tools on learning, broken down by field of study. The results reveal that student assessments vary significantly across different fields (Chi², p < 0.001, Cramer’s V = 0.17).

Figure 4

Students of computer and technological sciences most often indicate that AI has more positive than negative effects – more than 40% of students in these fields chose this answer. Students of physical and social sciences also tended to rate the impact as positive or balanced. In contrast, arts and educational sciences students were more cautious. Students of arts showed higher proportions indicating uncertainty or perceiving negative effects. Health science students were also quite skeptical, with a significant proportion of them stating that the impact of AI on learning is more negative (18%). Table 1 presents the survey data by type of educational institution and field of study.

The results show clear institutional and disciplinary differences. University students most often reported a balanced or positive view of AI’s impact, particularly in computer, social, and physical sciences. Educational sciences students also tended to view AI’s influence as balanced, though some expressed uncertainty. Health sciences students were comparatively more skeptical, indicating higher proportions of negative evaluations. Statistical analysis confirmed a significant relationship between among university students the field of study and the assessment of the impact of AI (Chi², p = 0.031).

Among college students, attitudes followed a similar pattern, though slightly more respondents viewed AI positively overall. Uncertainty was higher in colleges, especially in educational and technological fields. A statistically significant relationship between field of study and assessment of the impact of AI was also found among college students (Chi², p < 0.001).

Compared to university students, college students are more likely to rate the impact of AI positively (31.5% in colleges, 26.7% in universities), but the lack of knowledge about the impact of AI is higher in colleges (15.3% in colleges, 11.8% in universities). Universities have a more balanced view (44.4% in colleges, 46% in universities), but university students, especially in the field of educational sciences, have a slightly more positive view of the impact of AI. Statistically significant differences confirm that the field of study influences students’ attitudes towards the application of AI models to learning. Strategies for integrating AI in higher education therefore need to be tailored to the specificities and needs of particular fields of study.

The relationship between the use of AI tools and their impact on learning. The results of the study showed a moderate correlation between students’ use of AI tools in their studies and their attitudes toward the impact of these technologies on higher education (Chi², p < 0.001; Cramer’s V = 0.3). Students who actively use AI tools such as ChatGPT, code assistants, or AI-based learning platforms, were more likely to say that these tools have a more positive than negative impact on their studies (33.2%) compared to those who do not use these tools (13.1%). In addition, users were more likely to indicate that the impact of AI tools is balanced – neither entirely positive nor entirely negative (48.4%). Meanwhile, those who do not use AI tools were more likely to choose the answer ‘I don’t know’ (39.7%), which may indicate not only a lack of knowledge but also less exposure to the real possibilities of applying these technologies. These data suggest that practical experience with AI tools leads to a more favourable assessment of them, which is important when developing strategies to promote the targeted and responsible integration of AI into the study process.

4.3 Qualitative Results: Perceived Benefits and Challenges of AI in Learning

Respondents were asked open-ended questions in which they were asked to indicate the main positive and negative impacts of AI in an academic context. Qualitative content analysis was used to summarise and systematise the results. The responses were divided into thematic categories according to their dominant semantic meanings. Each response could be coded into one or more categories, as respondents often mentioned multiple themes. The results are summarised by institution type and main thematic categories.

5.1 Institutional and Disciplinary Differences

The study results indicate that university and college students use AI tools considerably more actively (78.8% and 75.5%, respectively) than vocational school students (46.5%). Insufficient knowledge about AI capabilities was identified as a primary barrier, particularly among vocational school students (14% were unaware of the possibility of using AI, compared with 2.2% in universities and 0.5% in colleges). Numerous studies confirm that knowledge gaps represent a major obstacle to AI tool implementation (DEC, 2024; Gesser-Edelsburg et al., 2024; Jereb & Urh, 2024).

Across all educational sectors, ChatGPT dominates AI usage, with mean usage scores of M = 4.31 (SD = 1.72) in universities, M = 4.42 (SD = 2.03) in colleges, and M = 3.45 (SD = 2.03) in vocational schools – corresponding to usage between ‘monthly’ and ‘several times per week’ in higher education institutions. Vocational school students showed limited overall engagement, with the lowest rate of open-ended responses (43.6%, versus 72% for universities and 55.6% for colleges). They primarily viewed AI as a source of basic information (30.7%) and learning assistance (22.7%), while more advanced applications – such as idea generation, information organisation, or personalised learning – were absent (0%).

The caution among vocational school students likely stems from several factors. First, insufficient AI literacy leads to greater feelings of uncertainty (28.5% of vocational students do not know what the impact was). Second, vocational training is often more oriented toward practical skills, in which AI application may be less obvious or insufficiently integrated into curricula.

Analysis by field of study revealed significant differences in AI adoption and evaluation among higher education students (Chi², p < 0.001, Cramér’s V = 0.17). Students of computer and technological sciences were were among the most likely to report that AI has more positive than negative effects (over 40% in both fields), alongside the highest usage rates (90.3% in computer science).

Educational sciences students demonstrated the lowest level of AI usage (66.7%). This relative reluctance may reflect the pedagogical community’s concerns about ethical and pedagogical risks (Divekar et al., 2024; Farooqi, Amanat & Awan, 2024; Gesser-Edelsburg et al., 2024; Kamali, Alpat & Bozkurt, 2024). Meanwhile, computer science and technology students use AI more actively and evaluate it more positively, which aligns with international research showing that technical subjects lead other fields in AI usage for learning purposes (Divekar et al., 2024; Dogan, Goru Dogan & Bozkurt, 2023; Ward et al., 2024; Zhang & Aslan, 2021).

Arts students exhibited the greatest uncertainty, with 33% selecting ‘don’t know’ regarding AI’s impact – the highest rate across all disciplines. This may indicate both limited exposure to AI in creative curricula and ambivalence toward AI’s place in human-centered creative processes.

Health sciences students were the most sceptical, with 33.3% of university students reporting that AI has more negative than positive effects – the highest negative rate across all fields. This reflects legitimate professional concerns. In an Egyptian study (N = 614), about 60% of medical students expressed doubts about ChatGPT’s accuracy on medical topics, and more than 71.3% of medical students worried about potential misuse in medical education (Abdelhafiz et al., 2025). A German study (N = 6,300) similarly found that health sciences students had the lowest AI usage rate (52.7%) among all disciplines (von Garrel & Mayer, 2023).

Finally, the integration of AI tools raises important equity issues. In the absence of clear guidelines and structured support, a divide can emerge, with strong students becoming even stronger through the use of AI, while weaker students risk falling further behind (Gesser-Edelsburg et al., 2024). The 32.3 percentage point gap in AI usage between vocational (46.5%) and university (78.8%) students represents a critical equity concern that demands urgent institutional attention. Moreover, equity issues extend beyond technical skills – they are crucial in socially sensitive fields such as social work, educational policy and medicine, where responsible and inclusive integration of AI is essential (Kubullek, Kumaç & Dogangün, 2024). The task of educational institutions is therefore not only to provide access to AI tools, but also to ensure equal opportunities for learning, consultation and ethical support to prevent the widening of achievement gaps and promote balanced digital readiness across all student groups.

5.2 Contradictions in Cognitive Impact

Open-ended questions revealed a striking paradox at the core of students’ AI experiences: the same technology is simultaneously viewed as a tool for enhancing creativity (10.9% mentioned idea generation and creativity) and as a primary cause of creativity decline (14.1%). This contradiction also extends to critical thinking, with 22.4% of respondents expressing concern about the loss of critical thinking and independence – the second most frequently mentioned negative impact after concerns about information reliability (26.1%). This dual evaluation emphasises AI’s dual impact – it can be either a benefit or a threat (Bozkurt & Bae, 2024; Divekar et al., 2024; Gesser-Edelsburg et al., 2024).

When used as an auxiliary tool for idea generation or for preliminary information processing, AI can accelerate the learning process and create space for deeper thinking (Chan & Hu, 2023; Gluoksnyte, 2025). However, the greatest concern involves overreliance on AI, which encourages laziness and suppresses cognitive effort and critical reasoning – skills essential for intellectual growth (Bae & Bozkurt, 2024; Bozkurt et al., 2024; Kamali, Alpat & Bozkurt, 2024). If AI is used as a substitute for independent work, there is a risk that students will lack sufficient cognitive depth to navigate new challenges independently (Bozkurt et al., 2024). Students themselves may become trained to use tools but lack the resilience to work without them (Bozkurt et al., 2024).

This duality underscores the need for pedagogical balance: integrating AI as a complement rather than a substitute for human thought (Fuchs & Aguilos, 2023; Ward et al., 2024). Educators should design learning tasks that promote reflection, critical evaluation, and creativity while leveraging AI’s efficiency benefits (Bozkurt et al., 2024).

5.3 Practical Experience, Ethics, and the Institutional Response

The study identified a strong correlation between actual AI usage and more positive attitudes toward its impact (Chi², p < 0.001; Cramér’s V = 0.3). Students who actively use AI tools were substantially more likely to evaluate their impact as positive (33.2% vs. 13.1% among non-users) or balanced (48.4% vs. 39.7% ‘don’t know’ among non-users). This pattern indicates that practical experience reduces uncertainty and shapes more favourable evaluation (Chan & Hu, 2023; Kubullek, Kumaç & Dogangün, 2024).

However, despite this growing familiarity, the results also reveal a considerable gap in AI literacy and institutional preparedness. Students frequently reported uncertainty about when and how AI tools should be appropriately used in their studies. This aligns with international findings which show that only 5% of students considered their institutions’ AI policies to be clear and comprehensive (DEC, 2024). In the Lithuanian context, this uncertainty was most pronounced among vocational school students, who showed both the lowest rates of AI use and the highest proportion of ‘don’t know’ responses (28.5%, compared with 11.8% in universities). This trend indicates not only a lack of systematic institutional guidance but also limited exposure to the practical potential and ethical implications of AI, reflecting broader differences in digital readiness across education levels.

The qualitative analysis also highlighted ethical issues directly related to the lack of AI literacy. Students mentioned risks related to plagiarism (12.5%), academic integrity and overreliance on AI-generated content (12.1% expressed concerns about technology dependency), demonstrating an awareness of ethical dilemmas but also of unclear institutional boundaries. These findings are consistent with broader debates about authorship and ownership in the age of GenAI (Bozkurt, 2024a; Xiao et al., 2025). When AI contributes to the creation of written works or ideas, questions arise about who owns the content created and to what extent AI’s involvement is ethically acceptable – questions that remain largely unresolved in current institutional systems.