6.1. Attitudes towards Chat GPT
The first section of the study explores respondents’ perceptions of the tool. Figure 1 presents data on students' enjoyment and interest in using the tool, its effectiveness in facilitating learning, and its impact on their learning performance. The results point to an overall positive disposition among students which reflects favorable perspectives across multiple dimensions. The data paints a compelling picture of the positive reception of Chat GPT among students in Tunisia, with a significant majority expressing both enjoyment (63%) and interest (66%) in interacting with the tool. These findings suggest that Chat GPT effectively delivers a pleasurable and engaging experience and fosters an environment conducive to active participation and sustained interaction. High levels of enjoyment indicate that users find the platform satisfying and enjoyable, likely leading to increased user retention and engagement over time. In a similar vein, Sengupta and Chakraborty (2020) noted that students generally perceive AI tools like ChatGPT positively, making learning more engaging and interactive. Along, similar lines, Alotaibi et al. (2020) found out that the use of AI in educational settings enhances student engagement and improves their learning outcomes.
The results underscore both the strengths of ChatGPT in meeting student expectations and the opportunities it offers to enhance their academic tasks. A significant 73.2% of respondents acknowledge that using ChatGPT significantly enhances the learning process, while 71% agree or strongly agree that it improves learning performance. These findings indicate that many students attribute positive academic outcomes to their interactions with ChatGPT. Mean values ranging from 3.65 to 3.94 on a scale of 1 to 5 affirm a generally positive attitude toward the tool, with slight variability indicated by standard deviations ranging from 0.91 to 0.98.
Students' positive attitudes toward ChatGPT reflect enjoyment, interest, and recognition of its facilitative role in learning and academic performance. Similar trends are observed in global studies, such as Sengupta and Chakraborty (2020), which highlight AI tools' capacity to enhance engagement and interactivity in learning. This universal appeal is reflected in the cultural context of Tunisia, where ChatGPT has garnered widespread interest for its ability to sustain user attention and foster autonomy in topic exploration. This aligns with insights from Nguyen et al. (2021), who emphasize the role of AI tools in promoting self-directed learning and exploration. Understanding the factors influencing varying attitudes among users, as indicated by standard deviations, could provide valuable insights for the continued development and implementation of ChatGPT in educational settings in Tunisia. These results were validated with qualitative data, which illustrate how ChatGPT meets students' needs and expectations by offering valuable support in their learning and decision-making processes. The tool's perceived intelligence, problem-solving capabilities, and user-friendliness contribute to its positive reception among students in Tunisia, as showcased through the quotes below.
Quote 3
Chat GPT is really smart enough to do comparisons and help me choose the best things when I have totally no idea about their specifications, it can calculate the probability of anything and give you 85% a good result, and I think it’s really incredible how they manage to develop such a powerful AI system that couldn’t stop increasing it data, its mind boggling (°◡°)
Quote 4: I had a very difficult case in which I was confused which one should I opt for (it was about two strategies) and thanks to ChatGPT, it gave me a lot of options for both strategies and a lot of supporting arguments. I finally got the right choice thanks to Chat GPT. So I guess it became my new favourite thing (+ emoji: grinning face with smiling eyes).
Quote 3 showcases ChatGPT's capabilities in providing comparative analyses and assisting users with complex decision-making. The student highlights the AI's proficiency in evaluating specifications, which indicates a reliance on its computational abilities. The mention of receiving an "85% good result" reflects a perceived trust in the system's output, suggesting that the user values its recommendations despite inherent uncertainties. The phrase "mind-boggling" conveys both awe and recognition of the rapid advancements in AI technology, underscoring the transformative impact of such tools on users' experiences. Quote 4 presents a personal narrative where the student faced a challenging decision between two strategies. The gratitude expressed towards ChatGPT illustrates the platform's role as a supportive resource in educational contexts. By providing multiple options and supporting arguments, ChatGPT not only facilitated the decision-making process but also empowered the student to arrive at an informed choice. The use of the grinning face with smiling eyes emoji further emphasizes the student's positive emotional response and satisfaction with the outcome. This indicates that ChatGPT is not merely a functional tool but has also become a trusted companion in the student's academic journey.
6.2. Chat GPT: Perceived usefulness and ease of use
The findings from Fig. 2 below offer insights into how university students in Tunisia perceive the usefulness and ease of use of Chat GPT. A significant majority of respondents (85%) view Chat GPT positively for its ability to quickly retrieve information, with 69% also endorsing its effectiveness in answering questions. However, opinions diverge regarding its ease of use: while a substantial percentage (85%) find it easy to use, a notable portion (40%) strongly agrees with this assessment. This variation likely stems from the novelty of the tool, as users are still exploring its capabilities and potential applications. As users become more familiar with Chat GPT, we anticipate a convergence of opinions and a clearer understanding of its usability. Examining task-specific ease, 61% of respondents agree or strongly agree with its effectiveness, while 23.4% express disagreement. Mean values ranging from 3.52 to 4.14 on a scale of 1 to 5 indicate generally positive perceptions of Chat GPT, particularly as a resource for answering questions (highest mean). Standard deviations ranging from 0.92 to 1.15 suggest moderate to slightly higher levels of dispersion in responses across assessed dimensions, underscoring the variability in students' opinions. It is crucial to acknowledge the dichotomy in perceptions, with students expressing both favorable and critical perspectives. The positive outlook on the utility of the Chat GPT aligns with its potential to enhance information retrieval and academic support. However, the diverse opinions on ease of use and task-specific functionality underscore the need for targeted interventions or improvements to address varying levels of user satisfaction. The examples below showcase the intricacies of dealing with ChatGPT.
Quote 5: c vrai que chatgpt nous aide trop mais je vois que ses réponses sont vagues pour moi par exemple pour avoir une réponse a qq chose je dois passer par plusieurs étapes pour lui expliquer et parfois j’ai pas la réponse que je veux parfois je trouve qu’il est un peu bête (French, sic). (English: It's true that ChatGPT helps us a lot, but I find its answers vague. For example, to get an answer to something, I have to go through several steps to explain, and sometimes I don't get the answer I want. Sometimes I find it a bit dumb).
Quote 6
Chat GPT is really unreliable if you don’t know how to prompt it. Giving a one-line command doesn’t get you anywhere. I use ChatGPT every day at work I write paragraphs explaining the context, the company, what exactly I want for ChatGPT to generate me ideas/ text.
The quotes above highlight the nuanced dynamics of engaging with ChatGPT. Quote 5 reflects the student's ambivalence towards ChatGPT. While they acknowledge its supportive role in their academic journey, they also encounter frustrations related to the vagueness of its responses. The need to go through several steps to articulate their queries indicates a struggle for clarity and precision, suggesting that the tool does not always meet their expectations. This duality highlights the inherent tension between recognizing the utility of ChatGPT and grappling with its limitations. Quote 6 reinforces this sentiment by emphasizing the importance of crafting effective prompts. The student's assertion that a simple, one-line command yields inadequate results underscores the need for specificity in communication with the AI. This requirement for detailed context illustrates that while ChatGPT can be a powerful resource, its effectiveness hinges on the user's ability to engage with it thoughtfully. Together, these quotes reveal a complex relationship where users must navigate both the advantages and challenges of utilizing ChatGPT, reflecting a broader trend in the integration of AI tools in education.
6.3. Chat GPT: Behavioral intentions
The results presented in Table 2 offer insights into the behavioral intentions of university students regarding the use of Chat GPT. The findings suggest an overall positive inclination toward future engagement with the platform, yet with notable variations in the anticipated frequency of usage. In terms of the intention to use Chat GPT more often in the future, a substantial 73% of the respondents expressed agreement or strong agreement. This positive inclination aligns with the notion that students perceive Chat GPT as a valuable tool, indicating a desire to integrate it further into their academic and daily tasks. Additionally, the mean value of 3.82 supports the interpretation of a generally positive behavioral intention. Similarly, the respondents exhibited a positive attitude toward recommending Chat GPT to others for their tasks or questions in the future, with 67.4% expressing agreement or strong agreement. This finding signifies a potential diffusion of Chat GPT usage within the academic community, as students are willing to advocate for its utility. The mean value of 3.71 further reinforces the overall positive sentiment in this aspect.
Table 2
Behavioral intention to use Chat GPT
| Behavioral intention to use Chat GPT | (1) | (2) | (3) | (4) | (5) | Mean | SD |
| I intend to use Chat GPT more often in the future | 21 3% | 58 8.6% | 101 15% | 343 51% | 152 22.5% | 3.82 | 0.97 |
| How likely are you to recommend Chat GPT to others for their tasks or questions in the future? | 41 6% | 37 5.5% | 144 21.4% | 310 46% | 141 21% | 3.71 | 0.97 |
| | (6) | (7) | (8) | (9) | (10) | Mean | SD |
| How often do you anticipate using Chat GPT in your daily or weekly tasks in the future? | 85 13% | 187 28% | 131 19% | 154 23% | 117 17% | 3.00 | 1.10 |
* (1) = very unlikely, (2) = unlikely, (3) = neutral, (4) = likely, (5) = very likely, (6) = rarely, (7) = occasionally, (8) = regularly (9) = frequently, (10) = always SD = standard deviation
The anticipated frequency of use presents a nuanced picture. While 45% of respondents expressed a proclivity toward regular or frequent usage in their daily or weekly tasks, 41% indicated a more reserved approach and reported a rare or occasional usage. The mean value of 3.00 reflects this diversity in responses. It suggests a less definitive stance among users regarding their future interaction with Chat GPT. This variation might be influenced by factors such as individual preferences, specific use cases, and the perceived effectiveness of Chat GPT in fulfilling diverse needs. The survey results indicate a generally positive behavioral intention to use Chat GPT in the future, both for personal use and recommendations to others. However, the variability in anticipated frequency underscores the importance of understanding the factors influencing this aspect and addressing any user concerns. Further qualitative research could uncover specific use cases, challenges, and opportunities, guiding strategies for enhancing user engagement and satisfaction. Understanding the specific contexts where Chat GPT proves most beneficial and addressing any potential limitations or challenges identified by users can guide the development of targeted interventions. Increasing awareness among students about the diverse functionalities and applications of Chat GPT could also boost its adoption. Thus, while there is a positive behavioral intention among Tunisian university students towards using Chat GPT, the variability in anticipated frequency calls for a deeper understanding of user preferences and needs, which can inform strategies for optimal integration and usability in the educational context of Tunisia.
6.4. Perceived credibility and privacy
Table 3 provides insights into the perceived credibility and social influence of Chat GPT among Tunisian university students. In terms of credibility, 68% of respondents agreed or strongly agreed that Chat GPT is a trustworthy source of information, with a mean value of 3.33. Additionally, 69.6% expressed confidence in the accuracy of the information provided by Chat GPT. This suggests consistent perceptions of its credibility. These findings indicate that a significant proportion of students rely on Chat GPT for accurate information, which could influence their learning habits and preferences for digital tools over traditional methods.
Regarding social influence, 59% of respondents reported starting to use Chat GPT for learning because they observed others using it in educational contexts, highlighting the impact of peer behavior on adoption. The mean value of 1.77 indicates a positive stance towards this influence. Furthermore, 84.4% believe that many people in their social network use Chat GPT regularly, with a mean value of 1.23. This underscores its widespread adoption within these circles. This demonstrates the powerful role of social networks in technology adoption. It also points out that peer recommendations and visible use can significantly boost the integration of Chat GPT into educational practices.
Table 3
Chat GPT: Perceived credibility and social influence
| Perceived credibility | *SD (1) | *D (2) | *N (3) | *A (4) | *SA (5) | Mean | *SD |
| ChatGPT is a trustworthy source of information | 37 (5.5%) | 142 (21%) | 264 (39%) | 194 (29%) | 36 (5.3%) | 3.33 | 1.02 |
| Chat GPT provides accurate information | 39 (5.8%) | 161 (24%) | 235 (35%) | 200 (29.6%) | 40 (6%) | 3.33 | 0.94 |
| Perceived credibility | *SD (1) | *D (2) | *N (3) | *A (4) | *SA (5) | Mean | *SD |
| Perceived social influence | YES | NO | PERHAPS | MEAN | SD |
| Have you started using ChatGPT for learning because you saw others using it in educational contexts? | 397 (59%) | 126 (19%) | 149 (22%) | 1.77 | 0.62 |
| Do you believe that a significant number of people in your social network are using ChatGPT regularly? | 569 (84.4%) | 23 (3.4) | 82 (12.2%) | 1.23 | 0.57 |
*SD = Strongly Disagree, D = Disagree, N = Neutral, A = Agree, SA = Strongly Agree, SD = Standard Deviation
These results prompt a discussion on the symbiotic relationship between perceived credibility and social influence. The moderately high level of trust in the credibility of Chat GPT, coupled with the acknowledgment of its widespread use in social networks, underscores the role of social influence in shaping perceptions and behaviors related to technology adoption. However, further exploration is warranted to understand the specific factors that contribute to the perceived credibility of Chat GPT and the dynamics of its adoption within social networks.
Table 4 provides insights into Tunisian university students' perceptions of the privacy and security of using Chat GPT. Regarding privacy concerns, approximately 59.6% of respondents expressed some level of apprehension, 33.7% were neutral, and 6.6% were strongly concerned. The mean value of 3.0 suggests a moderate level of concern among students about the privacy of their information when using Chat GPT. This finding underscores the need for the platform to address privacy-related apprehensions to ensure user trust and acceptance. By enhancing transparency about data practices and implementing robust privacy measures, Chat GPT can alleviate these concerns and build a stronger user base. On the other hand, respondents exhibited a relatively higher level of confidence in the security measures implemented by Chat GPT. A combined 61% agreed or strongly agreed that they were confident Chat GPT would keep their information secure, with a mean value of 3.29. This indicates a generally positive perception regarding the platform's ability to protect user data. These results suggest that while students are moderately concerned about privacy, they are reassured by the platform's security measures.
Table 4
Perceived privacy and security
| Perceived privacy and security | SD (1) | D (2) | N (3) | A (4) | SA (5) | Mean | SD |
| I am concerned about the privacy of my information when using Chat GPT | 44 6.6% | 149 22% | 226 33.7% | 175 26% | 76 11% | 3.0 | 1.02 |
| I am confident that Chat GPT will keep my information secure | 62 (9%) | 182 (27%) | 287 (43%) | 118 (18%) | 22 (3%) | 3.29 | 0.99 |
| Chat GPT takes adequate measures to protect my privacy | YES (1) | NO (2) | Maybe (3) | Mean | SD |
| 94 (14%) | 129 (19%) | 447 (67%) | 2.69 | 0.68 |
However, when asked if Chat GPT takes adequate measures to protect their privacy, the responses indicate a more varied perspective. While 67% of respondents expressed a neutral stance, a notable 33% either agreed or strongly agreed that Chat GPT took adequate measures. The mean value of 2.69 suggests a moderate level of confidence in the platform's privacy protection measures. These findings prompt a discussion on the delicate balance between user concerns about privacy and the perceived effectiveness of the security measures implemented by Chat GPT. Addressing these concerns through transparent communication about privacy practices, data encryption, and user control over their information becomes crucial to fostering a sense of trust among users. The moderate level of concern regarding privacy highlights the need for ongoing efforts by developers to enhance and effectively communicate security measures. Establishing a robust privacy framework that aligns with users' expectations is essential for contributing to a more positive user experience. By focusing on these areas, Chat GPT can build stronger user trust and ensure greater acceptance and satisfaction among its users.
6.5. CHAT GPT: The lived experience
The thematic content analysis of student responses to the question, "Can you share a specific experience where ChatGPT was particularly helpful or challenging for you in finding information or answering questions?" (24,000 words of raw text in Arabic, English, and French) revealed diverse perspectives clustered under rubrics as illustrated in Table 5 below. The analysis identified approximately 1,850 frequencies of mention of the subcategories. The most frequently mentioned categories are academic and professional use and general knowledge and coding, both at 27%. Students appreciate ChatGPT's support for assignments, courses, summaries, and research assistance, underscoring its potential as a valuable educational tool. Positive outcomes, such as recommendations, ideas for presentations, and C# programming practice, appear in 13% of mentions, suggesting that ChatGPT enhances creativity and problem-solving skills. However, 11% of mentions highlight issues with confusing or incorrect responses, indicating a need for improved answer accuracy. Neutral or mixed experiences, including irregular use and inconsistent results, account for another 11%, indicating varied engagement levels and a need for better support for non-English speakers. Ease of use and efficiency are appreciated in 6% of mentions, while 7% highlight using ChatGPT for homework and research projects, indicating its role in facilitating academic tasks. Concerns about dependence on ChatGPT appear in 3% of feedback, highlighting the importance of encouraging critical thinking and using AI tools as supplements rather than replacements for learning. Language and communication issues, particularly difficulties with handling Arabic, make up 6.5% of mentions, emphasizing the need for improved language capabilities. Lastly, 10% of responses address accuracy and precision issues in specific topics, suggesting that while ChatGPT is useful, it requires cautious use for subjects that demand high accuracy.
Table 5
Thematic content analysis
| Categories | Examples of Subcategories | Frequency |
| Academic and Professional Use | Support (e.g., assignments, courses), Summarizing, Research Assistance | 27% |
| General Knowledge and Coding | Coding assistance, general knowledge inquiries | 27% |
| Confusing or Incorrect Responses | Incorrect answers (e.g., homework, standards), different answers with "Are You Sure?" | 11% |
| Positive Outcomes | Recommendations, ideas for presentations, C# Programming practice | 13% |
| Neutral or Mixed Experiences | Irregular use, inconsistent results | 11% |
| Ease of Use and Efficiency | General ease of use, availability of information | 6% |
| Homework and Research Use | Using ChatGPT for homework, using ChatGPT in projects and research | 7% |
| Dependence, Ethics, and Caution | Concerns about dependence on ChatGPT | 3% |
| Language and Communication | Clear and detailed explanations, difficulties in handling Arabic Language | 6.5% |
| Recommendations and Usage Diversity | Recommending ChatGPT to friends, diverse usage (e.g., projects, research) | 6% |
| Accuracy and Precision | Accuracy issues in specific topics, precision in giving information, confusing or incorrect responses | 10% |
Overall, the findings indicate that while ChatGPT offers substantial benefits, some areas need attention to enhance its effectiveness and reliability for students. Addressing the highlighted issues can significantly improve the user experience and educational outcomes. For instance, the 11% of mentions concerning confusing or incorrect responses emphasize the need for ongoing improvements in the accuracy of ChatGPT's answers. Ensuring that the AI provides reliable and consistent information is crucial for maintaining trust and encouraging its use in academic contexts. Feedback on language and communication issues, particularly difficulties with handling Arabic, points to a critical area for development. Enhancing ChatGPT's capabilities in non-English languages can make it more accessible and useful for a broader range of students. This improvement would meet the needs of those who currently find the tool less supportive due to language barriers.
The concerns about dependence on ChatGPT, raised by 3% of respondents, highlight an important aspect of AI integration in education. It is essential to promote critical thinking and ensure that students use ChatGPT as a supplement rather than a replacement for their learning efforts. Educators and developers should work together to create guidelines and educational strategies that balance AI assistance with the development of independent problem-solving skills. Moreover, the 10% of responses discussing accuracy and precision issues in specific topics suggest a need for careful consideration when using ChatGPT for subjects requiring high accuracy. Students and educators should be aware of these limitations and use the tool in conjunction with other reliable resources to ensure the quality of learning outcomes.
Students express mixed feelings toward ChatGPT, recognizing its utility in project preparation and quick information retrieval while also cautioning against over-reliance. One student affirms, "It helped me a lot to find information in a faster way," but advises, "It should be used smartly because using artificial intelligence is not allowed." This sentiment is echoed by another student who notes, "ChatGPT is a starting point but not the whole point."The idea of ChatGPT as a "starting point" is crucial to understanding its role in academic work. Students appreciate ChatGPT's ability to quickly provide a broad overview and initial information, which can save time and offer a foundation for further research. However, they also recognize the necessity of deeper investigation, cross-referencing multiple sources, and critical analysis to ensure the accuracy and comprehensiveness of their work. The following quotes from the corpus illustrate their perspectives:
Quote 6
Chat GPT is a start point but not the whole point. It is helpful with general knowledge and coding, however very inaccurate when it comes to numbers and calculations."
Quote 7: C'est vrai que ChatGPT nous aide trop, mais je trouve que ses réponses sont vagues (English: It is true that ChatGPT helps us a lot, but I find its answers vague).
These examples employ concessions and contrasts to present a balanced view of using ChatGPT. In Quote 6, the student emphasizes the necessity of human oversight, particularly in precision-dependent areas like numerical data and calculations. Quote 7 shows a student who recognizes the importance of interpreting ChatGPT feedback cautiously. Students acknowledge the value of ChatGPT as a tool for gathering initial information and generating ideas while also highlighting its limitations. They understand that ChatGPT should complement, not replace, traditional research methods and critical thinking skills. This balanced perspective underscores the importance of using ChatGPT as a supplementary tool while maintaining a critical approach to its outputs and recognizing its inherent limitations.
While the model proves helpful in general knowledge and coding tasks, a significant limitation arises when handling numbers and calculations. Many users report that ChatGPT is often inaccurate regarding numerical data. This highlights a critical area where the tool's performance falls short, especially in fields requiring precise numerical accuracy. Additionally, concerns exist about the vagueness and insufficiency of responses. Students express frustration when answers do not adequately address their questions or provide the needed detail. These observations reflect a nuanced understanding of ChatGPT's strengths and weaknesses. While it excels in tasks like general knowledge and coding assistance, its reliability diminishes in contexts requiring precise numerical data or detailed explanations. Users must approach its use with awareness of these limitations and supplement its outputs with additional research and critical analysis to ensure completeness and accuracy in their work. Ethical considerations also frequently arise in discussions about ChatGPT, particularly regarding its use in academic settings. Some users acknowledge the tool's usefulness for assignments, recognizing that it can replace tutors while also facilitating academic dishonesty. This underscores the importance of promoting responsible use of AI in education to prevent unethical practices while maximizing its educational benefits. An illustrative quote of this concern is as follows:
Quote 8
Personally I used Chat GPT, it is extremely useful for assignments. Not just because it answers all your questions that you ask, but it completely destroys the use of tutors. However, it should be noted that it can be used productively but unethically because it is easier to cheat and just copy whatever the AI is providing (sic).
The quote above encapsulates a recurring ethical concern associated with the use of ChatGPT in academic contexts. While the tool offers substantial benefits, such as efficient assignment completion and reduced reliance on human tutors; it also raises ethical considerations. The user acknowledges that ChatGPT can be misused for academic dishonesty. This misuse enables students to bypass genuine learning and simply copy answers generated by the AI. This ethical dilemma highlights the need for responsible use of AI tools in education, where maintaining academic integrity is paramount. Educational institutions must navigate these challenges by implementing clear guidelines and promoting ethical behavior among students. Strategies may include educating students about the appropriate use of AI tools, emphasizing critical thinking and originality in assignments, and integrating AI literacy into academic curricula. By fostering a culture of academic honesty and responsible technology use, institutions can harness the benefits of AI while mitigating its potential for unethical misuse.
This nuanced feedback highlights both the potential and limitations of ChatGPT. It emphasizes the importance of thoughtful and discerning usage in academic endeavors. Students express mixed feelings that provide valuable insights into their experiences. ChatGPT significantly aids in project preparation by swiftly finding information and enhancing research efficiency. There is a recognition of the need to assess and supplement AI-generated information with human judgment, addressing ethical considerations. While it serves as a starting point in research, ChatGPT cannot replace human understanding and analysis. It receives praise for aiding general knowledge and coding but faces criticism for inaccuracies in numbers and calculations, suggesting areas for improvement. Vague responses and absolute answers reveal limitations in providing precise information, urging developers to enhance contextual understanding and accuracy. Feedback in multiple languages showcases ChatGPT's multilingual capabilities, though concerns arise about its depth and accuracy in less common languages. Overall, addressing these insights through continuous development will be essential to optimize ChatGPT's effectiveness across academic contexts.