Generative AI in Higher Education: A Student Perspective

Research Paper Recommended citation: Uukkivi, A., Retsnoi, V., Vilms, M., Petjärv, B., Labanova, O., & Safiulina, E. (2025). Generative AI in Higher Education: A Student Perspective. In Kangaslampi, R., Langie, G., Järvinen, H.-M., & Nagy, B. (Eds.), SEFI 53rd Annual Conference. European Society for Engineering Education (SEFI), Tampere, Finland. DOI: 10.5281/zenodo.17631754. This Conference Paper is brought to you for open access by the 53rd Annual Conference of the European Society for Engineering Education (SEFI) at Tampere University in Tampere, Finland. This work is licensed under a Creative Commons Attribution-NonCommercial-Share Alike 4.0 International License. GENERATIVE AI IN HIGHER EDUCATION: A SHORT SURVEY OF STUDENT PERSPECTIVES A. Uukkivi a, 1 , V. Retšnoi b, M. Vilms b, B. Petjärv b, O. Labanova b, E. Safiulina b, a TTK University of Applied Sciences, Tallinn, Estonia, ORCID 0000-0002-9490-4849 b TTK University of Applied Sciences, Tallinn, Estonia, ORCID Conference Key Areas: Digital tools and AI in engineering education, Improving higher engineering education through researching engineering education Keywords: Challenges of AI in education, Ethics and AI use, Critical thinking and AI, AI guidelines in universities, AI adoption in education ABSTRACT The integration of generative artificial intelligence (AI) tools into higher education is rapidly transforming the learning experience. This study investigates university students' experiences and perceptions of generative AI use in their studies. A survey was conducted to explore how AI tools are utilized, what attitudes students have toward them, what concerns and expectations they express, and what impact AI has on learning. Findings indicate that most students use generative AI chatbots for brainstorming, summarizing, and simplifying study materials, while fewer rely on it for completing assignments or preparing for exams. Although AI is generally perceived as useful and easy to use, issues such as misinformation, lack of proper citations, and potential negative effects on critical thinking were frequently mentioned. Also, concerns have been raised about potential overreliance and the ethical implications of AI-generated content. The analysis also revealed that younger students and those with more experience using AI tend to evaluate its usefulness more highly and students who demonstrate a greater willingness to use generative AI tools in learning tend to rate their knowledge and skills in using AI for learning more highly. As universities navigate on rapidly evolving landscape, there is a pressing need for academically rigorous and evidence-based insights to guide policies and practices. The findings inform educators in developing guidelines for responsible AI use, ensuring that students benefit from AI-driven tools while maintaining academic integrity and critical thinking skills. 1 Corresponding Author A. Uukkivi Anne.uukkiv[email protected] 1 INTRODUCTION AI is increasingly being integrated into higher education, transforming learning and teaching (Chen et al., 2024; Kizilcec et al., 2024). Students' experiences and opinions regarding the use of AI are diverse, as they recognize its benefits and challenges while also establishing expectations for the future (Babu et al., 2024; Chen et al., 2024). Students utilize AI-based tools for various purposes to enhance their learning and efficiency (Chen et al., 2024). For personalized learning support, AI-powered chatbots and virtual assistants provide educational materials and instant feedback, which is especially valuable for language acquisition and understanding complex topics (Chen et al., 2024; Essel et al., 2022). AI is used in writing to generate ideas, improve grammar and style, and create coherent text (Chen et al., 2024; Darvishi et al., 2024). AI is also considered useful for research, assisting in data analysis, generating hypotheses, and identifying connections that contribute to the quality of academic work (Chen et al., 2024; Garg et al., 2025). Additionally, AI enables students to automate group tasks and create study materials, saving time and improving the learning process (Chen et al., 2024; Darvishi et al., 2024). Students' attitudes toward AI are generally positive, particularly regarding its potential to make learning more efficient and accessible (Chen et al., 2024; Rodway & Schepman, 2023). However, it is also important to consider potential issues. Students value AI as a tool that helps them better understand complex topics, receive quick feedback, and personalize their learning experience according to their needs (Chen et al., 2024; Hew et al., 2022; Rodway & Schepman, 2023). Several positive effects of AI in education have been identified. AI allows students to quickly access relevant information and automate routine tasks, enabling them to focus on more important topics and increasing learning efficiency (Chen et al., 2024; Crompton & Burke, 2023). AI also helps tailor learning materials and methods to students' individual needs and learning styles (Chen et al., 2024; Hew et al., 2022; Lee et al., 2024). AI-based systems can provide fast and comprehensive feedback, helping students correct mistakes and learn more effectively (Chen et al., 2024; Crompton & Burke, 2023). Despite the many advantages, students have concerns regarding AI in education. They worry that AI might encourage plagiarism and reduce their engagement in the learning process (Chan, 2023; Chen et al., 2024). AI systems collect and analyze large amounts of data, raising privacy and data security concerns (Chan & Hu, 2023; Chen et al., 2024). The use of AI in education also brings ethical questions, particularly regarding fairness, equality, and accountability (Chen et al., 2024; Kizilcec et al., 2024). Students expect AI integration in higher education to enhance the learning process and improve access to information (Babu et al., 2024; Chen et al., 2024). They see AI's potential in automating routine tasks and providing personalized support to improve their learning experience (Chen et al., 2024; Crompton & Burke, 2023). However, it is crucial for educators and policymakers to develop guidelines that ensure the responsible use of AI while maintaining academic integrity and essential cognitive skills (Chen et al., 2024; Kizilcec et al., 2024). Building on this context, the present study is guided by the following hypotheses: H1: Students primarily use generative AI to simplify study materials and find it useful, but express concerns about misinformation and its impact on critical thinking. H2: Younger students and those with more experience using AI tend to rate its usefulness higher and are more willing to use it in their studies. These hypotheses aim to deepen understanding of how students engage with generative AI, their perceptions of its benefits and risks, and the factors influencing their attitudes and willingness to adopt AI tools in higher education. 2 METHODOLOGY A survey was conducted among the students to investigate students' experiences and opinions regarding using AI in learning. Specifically, the survey seeks to understand how various AI tools and applications are utilized in learning, students' attitudes toward AI, and their impact on students’ educational experiences. Additionally, the study explores students' concerns and expectations about the integration of AI in learning. 2.1 Sample In the 2024/25 academic year, the total number of students at TTK University of Applied Sciences 2 (TTK UAS) was approximately 2760. The study group consisted of 518 students who responded to the survey. Thus, the response rate in the population was approximately 19%. Participants were categorized as follows: • By gender: females (55.79%) and males (44.21%). • By age: up to 20 years old (23.55%), 21–25 years old (38.22%), 26–30 years old (9.27%), 31–35 years old (9.07%), 36–40 years old (8.69%), and 41 years or older (11.20%). • By study form: daytime studies (51.54%), session-based studies (47.10%), and external studies (1.35%). • By institution: Institute of Service Economy (24.71%), Institute of Logistics (20.27%), Institute of Engineering and Circular Economy (17.37%), Institute of Technology (16.02%), Institute of Civil Engineering (15.64%), and Institute of Architecture (5.98%). • By year of study: first-year students (40.54%), second-year students (26.25%), third-year students (21.24%), and fourth-year or over nominal study time students (11.97%). Among the 518 students surveyed, it was found that 95 (18.34%) students had never used generative AI tools in their studies. Therefore, their responses were excluded from further analysis, and the total number of respondents was reduced to 423. 2.2 Data collection The data for the current study were collected in September and October 2024 through an online survey administered via Google Forms, facilitating easy access and participation for all students. The survey was designed to be self-administered, allowing respondents to fill out the questionnaire independently. This approach facilitated anonymity and encouraged honest responses, which is critical for obtaining accurate and reliable data. The survey consisted of seven closed-ended questions, focusing on students' use of generative AI tools in their learning. It explored the duration, frequency, and purposes of their AI use. Additionally, it examined the challenges students encountered while using AI and their evaluation of its effectiveness in learning. The survey included the following multiple-choice questions: Q1: How long have you been using AI for learning? 2 TTK University of Applied Sciences (TTK UAS) is the largest university of applied sciences in Estonia. In Estonian, TTK is the abbreviation for Tallinna Tehnikakõrgkool. Less than one year, one year, two years, more than two years. Q2: How often do you use AI for learning? A few times per semester, a few times per month, a few times per week, every day. Q3: How would you rate your knowledge and skills in using AI for learning? ‘No knowledge’– 1…5 – ‘Expert knowledge’. Q4: Which generative AI tools do you use for learning? Text bots, image generators, audio generators, video generators. Q5: What do you use generative AI tools for in your studies? 17 options to choose from, including idea generation, providing examples, rewording text, and so on. Q6: A self-report questionnaire on students' evaluation of generative AI use, consisting of eleven items presented in Table 2. Q7: What problems do you see with the use of generative AI in education? 11 problems to choose from, including incorrect or misleading answers, over-reliance on AI, AI inaccuracies, and so on. Furthermore, the survey contained two open-ended questions to explore students’ expectations for improvements in AI use within the university’s learning environment and the type of support or training they require to effectively integrate AI into their studies: Q8: What improvements would you like to see in the use of AI in higher education? Q9: What kind of support or training would help you use AI effectively in your studies? 2.3 Data analysis First, the descriptive statistics for the relevant variables were calculated for the quantitative questions. Next, the relationships between these variables were analyzed. Finally, it was examined whether these variables differed significantly according to independent factors such as gender, age, study form, institution, and year of study. Data were analyzed using MS Excel and the statistical software R. Correlations between variables were examined using Spearman’s rho coefficient. Differences between groups were assessed using the Wilcoxon rank-sum test and the KruskalWallis rank-sum test. Significance was set at a minimum level of 0.05, with other significance levels (0.01 and 0.001) also reported. The open-ended responses were analyzed using thematic analysis, where recurring themes and patterns were identified through systematic coding and categorization of students' answers. The responses were analyzed inductively by identifying recurring patterns and themes. 3 RESULTS In response to the question about the duration of AI use (Q1), the participants answered as follows: less than one year – 44.21%, one year – 25.30%, two years – 19.15%, and more than two years – 11.35%. A significant majority of participants (69.51%) have been using AI for learning for one year or less, suggesting that AI is a relatively new tool for most learners. In response to the question about the frequency of AI use (Q2), the participants answered as follows: a few times per semester – 34.75%, a few times per month – 31.91%, a few times per week – 27.42%, and every day – 5.91%. A significant majority of participants use AI for learning only occasionally, with 66.66% reporting usage just a few times per month or semester. Participants were also asked to rate their knowledge and skills in using AI for learning (Q3) on a 5-point Likert scale: 1 – ‘No knowledge’ (0.71%), 2 – ‘Limited knowledge’ (20.09%), 3 – ‘Average knowledge’ (43.74%), 4 – ‘Good knowledge’ (28.37%), and 5 – ‘Expert knowledge’ (7.09%). The results indicated that the participants rated their knowledge and skills in using AI for learning as average (M = 3.21, SD = 0.87). Regarding the question about which generative AI tools students typically use in their learning (Q4), many listed multiple tools. The results indicate a strong preference for text-based AI tools, which were used by 96.22% of respondents. This was followed by image generators (24.35%), while audio and video generators were used far less frequently (both at 3.78%). Regarding the question about the purposes generative AI tools typically serve in students' learning (Q5), many respondents selected multiple tools. The most common uses included idea generation (74%), providing examples (56.97%), rewording text (54.37%), and explaining concepts (53.43%). AI was also used for simplifying subjects (47.04%), summarizing material (44.21%), getting an overview of the topic (43.97%), checking spelling (40.90%), translating (39.72%), and solving problems (maths, etc.) (31.91%). These results indicated that students used AI for a wide range of cognitive support tasks, with a strong emphasis on content understanding, language refinement, and idea development. In this study, a self-report questionnaire (Q6) was used to assess students' use of generative AI technologies. The ‘Evaluation of Generative AI Use’ scale consisted of eleven items (Table 2) measured on a 5-point Likert scale: 1 – ‘Never’, 2 – ‘Occasionally’, 3 – ‘Neutral’, 4 – ‘Mostly’, and 5 – ‘Always.’ Responses to the individual items were summed to calculate a total score, ranging from 15 to 55, with higher scores reflecting a greater willingness to use generative AI in learning. The scores of the items had satisfactory internal consistency reliability (Cronbach’s alpha 0.87). Descriptive statistics for the relevant scale variable and items were calculated and are presented in Table 1 and 2. Table 1. Descriptive Statistics 3 of the ‘Evaluation of Generative AI Use’ scale (n=423) Variable M SD MIN 25% 50% 75% MAX Skew Kurt Evaluation of GenAI Use 34.02 8.69 15.00 28.00 33.00 40.00 55.00 0.35 -0.36 Table 2. Means (M), standard deviations (SD), and conclusions for the ‘Evaluation of Generative AI Use’ items (n=423) Please evaluate the use of AI M SD Conclusion AI is useful for improving learning 3,56 1,09 Rather Mostly AI makes learning more interesting 3,05 1,22 Neutral, disagreements AI helps me better understand what I am learning 3,49 1,21 Rather Neutral AI helps me solve tasks faster 3,30 1,28 Neutral, disagreements AI helps me prepare better for exams and tests 2,84 1,26 Neutral, disagreements AI motivates me to study 2,32 1,27 Occasionally AI makes me more confident in my studies 2,77 1,36 Neutral, disagreements 3 M – mean, SD – sample standard deviation, MIN – the smallest value, MAX – the highest value, 25% –, 50% –, 75% – the 25th, 50th and 75th percentile, respectively, Skew – sample skewness, Kurt – sample kurtosis. Please evaluate the use of AI M SD Conclusion AI has a positive impact on my learning outcomes 3,19 1,18 Neutral, disagreements AI is a reliable source 2,40 1,07 Rather Occasionally AI is easy to use 3,88 1,14 Mostly In the future, I will use AI more in my learning 3,21 1,17 Neutral, disagreements In this study, the main purpose of the ‘Evaluation of Generative AI Use’ scale was to examine its correlations with other measured variables. In Table 1 the mean score (M) of 34.02 with a standard deviation (SD) of 8.69 indicated moderate overall willingness among students to use generative AI for learning. The distribution was approximately symmetrical, with a slight positive skew (Skew = 0.35) and modest platykurtosis (Kurt = -0.36), suggesting no major deviations from normality. In this survey, there was a question about the problems students perceive with the use of AI in learning (Q7). Students identified several challenges in using AI for learning, with the most common being incorrect or misleading answers (85.82%), over-reliance on AI (65.72%), AI inaccuracies (64.30%), lack of references (60.52%), and data security concerns (55.32%). Other issues included the need for human intelligence in working with AI (54.37%), its impact on critical thinking (45.39%), copyright concerns (43.97%), and accessibility challenges (33.10%). Spearman’s rho correlation analysis revealed a moderate, positive, and significant relationship between the ‘Evaluation of Generative AI Use’ scale and the responses to the question Q3: ‘How would you rate your knowledge and skills in using AI for learning?’ (ρ = 0.31, p < 0.001). This suggests that students who showed a greater willingness to use generative AI tools in learning tended to rate their own knowledge and skills in using AI for learning more highly. The Wilcoxon rank-sum test and the Kruskal-Wallis rank-sum test were used to assess statistically significant differences between groups based on gender, age, study form, institution, and year of study. It was found that there were no statistically significant differences between the groups based on gender, study form and institution with respect to all relevant variables. The Kruskal-Wallis rank-sum test revealed significant differences in the total scores for ‘Evaluation of Generative AI Use’ based on students’ age (χ2 = 18.314, df = 5, p < 0.01), duration of AI use (χ2 = 30.418, df = 3, p < 0.001), and frequency of AI use (χ2 = 73.962, df = 3, p < 0.001). Specifically, the test results showed that younger students, as well as those who used AI for longer periods of time or more frequently, scored higher in the ‘Evaluation of Generative AI Use’ compared to older students and those who used AI for shorter periods or less frequently. The last two survey questions were open-ended, allowing students to share their thoughts on AI use in their studies. The first question (Q8) asked what improvements they would like to see in AI integration at TTK. The analysis revealed three main themes: the need for clearer guidelines, concerns about the quality of AI-generated content, and issues with accessibility. Many students emphasized the importance of having clear institutional guidelines on how to use AI in learning. There was a strong preference for specific instructions tailored to different subjects, with practical examples to help students apply AI tools effectively. Concerns about the accuracy and reliability of AI-generated content were also frequently mentioned. Students highlighted the need for factually correct, up-to- date, and subject-specific AI outputs. Additionally, they pointed out that AI tools often lack proper citation systems, making it difficult to use them in academic work. Accessibility was another key concern. Several students reported difficulties accessing premium AI tools and suggested that the university should provide access to paid AI versions, create a central AI platform, and ensure that all students have equal opportunities to use these technologies. The second open-ended question (Q9) focused on the kind of support and training students need for effective AI use. Three main areas stood out: technical skills, critical evaluation skills, and ethical considerations. Many students expressed a need for training on writing better prompts, requesting hands-on exercises, subject-specific examples, and clear guides on using AI tools. Others highlighted the importance of developing their ability to critically assess AI-generated content, particularly in checking facts, evaluating source credibility, and validating AI-generated results. Ethical concerns were also frequently raised, with students asking for clearer guidance on citation rules, copyright issues, and data privacy when using AI in their studies. 4 DISCUSSION AND CONCLUSIONS The findings of this study underscore the increasing role of generative AI tools in higher education, though challenges and concerns remain. The analysis of the data revealed that approximately 18% of TTK UAS students had never used generative AI tools in their studies. Among students who use AI for learning, around 70% of students have been using AI tools for up to one year. The survey results reveal that while a significant portion of students actively use AI for learning, its usage remains largely text-based and primarily supports creativity and comprehension. These findings align with prior research indicating that students frequently use AI for summarization, brainstorming, and rewording text rather than for high-stakes academic tasks (Chen et al., 2024; Rodway & Schepman, 2023). Furthermore, the results also confirm earlier studies that highlight the potential of AI to assist in learning by providing quick access to information, simplifying complex topics, and facilitating personalized learning (Crompton & Burke, 2023; Lee et al., 2024). Despite the benefits, students expressed concerns about the reliability of AIgenerated content, with 85.82% citing incorrect or misleading answers as a major issue. This concern aligns with previous research, which has pointed out that AI models can produce hallucinated or biased responses, making them unreliable for academic use (Chan, 2023; Kizilcec et al., 2024). Additionally, over-reliance on AI emerged as a significant issue (65.72%), echoing concerns from prior studies that warn of AI potentially diminishing students' critical thinking and engagement in the learning process (Chan & Hu, 2023; Hew et al., 2022). The results also indicate that students who use AI more frequently tend to perceive themselves as more competent in using these tools. This pattern is expected, as previous studies have shown that digital literacy and AI familiarity significantly influence students’ confidence and willingness to adopt AI in education (Essel et al., 2022; Garg et al., 2025). Additionally, the study found that younger students and those with more frequent AI exposure rated AI’s usefulness higher, suggesting that AI adoption in education may continue to grow as students become more accustomed to these tools over time (Darvishi et al., 2024). Interestingly, the study highlights a divergence from some earlier findings regarding AI's motivational impact on students. While previous research suggests that AI can enhance motivation and engagement in learning (Hew et al., 2022), the present study found that students only occasionally felt motivated by AI (M = 2.32, SD = 1.27). This discrepancy may be due to the specific ways students interact with AI, using it as a tool for efficiency rather than engagement. Another notable finding is that students did not strongly agree that AI helps them solve tasks faster (M = 3.30, SD = 1.28), with responses showing neutrality and disagreement. This suggests that while AI is often seen as a tool for enhancing creativity and generating ideas, it may not always provide immediate efficiency in task completion. Prior research indicates that AI-generated outputs often require critical evaluation and refinement before being useful (Chen et al., 2024; Lee et al., 2024), which could explain why students do not perceive AI as a time-saving tool in every situation. Additionally, difficulties in formulating effective prompts (as reported by 29.31% of students) may contribute to the perception that AI does not significantly speed up their workflow. For TTK UAS, one of the key contributions of this study is the emphasis on institutional guidelines and accessibility as crucial factors for AI integration in education. Many students expressed the need for clearer AI usage policies, as well as improved access to high-quality AI tools. These findings support the argument that higher education institutions, including TTK UAS, play a vital role in establishing frameworks for responsible AI use (Kizilcec et al., 2024; Lee et al., 2024). Based on these insights, this study recommends the development of an institutional framework for AI use, incorporating guidelines, enhanced faculty training, and structured student support. Specifically, the following recommendations can be made: (1) create an institutional framework, including AI usage guidelines and access to high-quality AI tools; (2) develop teaching methodologies by integrating AI awareness into curricula and enhancing faculty skills in guiding AI usage; and (3) establish a support network through training and sharing best practices. These results provide a solid foundation for developing AI usage at the university, helping to make informed decisions about improving teaching methods and support systems. By implementing these measures, higher education institutions can ensure that students benefit from AI-driven tools while upholding academic integrity and critical thinking skills. REFERENCES Babu, Md. A., Yusuf, K. Md., Eni, L. N., Shekh, Md. S. J., & Sharmin, Mst. R. (2024). ChatGPT and generation ‘Z’: A study on the usage rates of ChatGPT. Social Sciences & Humanities Open, 10(101163). https://doi.org/10.1016/j.ssaho.2024.101163 Chan, C. K. Y. (2023). Is AI Changing the Rules of Academic Misconduct? An Indepth Look at Students’ Perceptions of ‘AI-giarism.’ Computer Science > Computers and Society. https://doi.org/10.48550/arXiv.2306.03358 Chan, C. K. Y., & Hu, W. (2023). Students’ voices on generative AI: perceptions, benefits, and challenges in higher education. International Journal of Educational Technology in Higher Education, 20(43). Chen, D., Liu, W., & Liu, X. (2024). What drives college students to use AI for L2 learning? Modeling the roles of self-efficacy, anxiety, and attitude based on an extended technology acceptance model. Acta Psychologica, 249. https://doi.org/10.1016/j.actpsy.2024.104442