Introduction

Artificial Intelligence (AI) chatbots have seen a meteoric rise in popularity over the past few years since they provide users with a novel and time-saving method of communication [1]. AI Chatbots are becoming increasingly sophisticated as a result of developments in artificial intelligence and natural language processing. They are currently being used in a variety of applications, including customer service, healthcare, and education [2].

ChatGPT-3.5, an advanced language model developed by OpenAI, has gained significant attention and popularity in recent years [3]. It is a generative AI that can generate text, images, code, and more. Through its extensive training on social media, websites, articles, datasets, books, and other internet texts, it can answer questions and help humans with tasks [4]. It has reached an estimated 100 million active users per month. However, the exact number of active users is unknown. Brandl [5] estimated the number to be approximately 100 million, if 10% of their website visitors sign up for the free service. ChatGPT was trained on a 300-billion-word dataset, which is approximately 570GB [6]. ChatGPT-3.5 was launched as a prototype on November 30, 2022, and was praised for its detailed and articulate responses across many fields [5]. ChatGPT-4, an updated version, launched in March 2023. This version is only available to ChatGPT Plus users or API developers [6].

The emergence of ChatGPT represents a significant advancement in natural language processing and AI [4]. With its impressive ability to generate coherent and contextually relevant responses, ChatGPT has found utility in various domains. It has emerged as a transformative tool in education, offering personalized learning experiences, efficient academic support, and opportunities for skill development. Its integration into educational settings has garnered significant attention due to its ability to streamline complex academic processes and foster interactive learning environments.

ChatGPT facilitates tailored learning experiences by adapting to individual student needs. It provides personalized feedback, automates assessments, and supports independent learning, making it an essential tool for modern education systems. For example, in programming education, ChatGPT has been shown to enhance computational thinking and algorithmic skills while improving creativity and problem-solving abilities. Similarly, in nursing education, ChatGPT has significantly boosted academic performance and research capabilities [7,8].

ChatGPT offers quick responses to queries, saving time for students and faculty alike. Its ability to synthesize information into concise formats reduces the need for extensive research across multiple sources. Additionally, it aids in language skill development by providing grammar corrections and vocabulary suggestions. However, concerns about accuracy persist, as ChatGPT sometimes generates false references or inconsistent outputs [9,10].

Recent studies highlight ChatGPT’s capacity to motivate students by fostering active learning processes. It improves engagement through interactive dialogues and tailored coaching, which are particularly beneficial in disciplines like English language learning [8,11]. Moreover, its use in creating collaborative environments has been linked to enhanced psychological well-being among students [9].

Despite its advantages, ethical challenges such as plagiarism and over-reliance on AI tools remain prevalent. Students often employ ChatGPT to complete assignments or exams without proper attribution, raising concerns about academic integrity. Educators emphasize the need for guidelines to ensure responsible usage while promoting critical thinking skills [12,13].

University students, in particular, have shown a keen interest in using ChatGPT for a range of academic and personal purposes. It can understand and respond to natural language queries, making it an ideal tool for education [14]. The use of ChatGPT in education can have a significant impact on student learning, as it can provide personalized assistance to students and help them overcome their learning challenges [15]. University students use ChatGPT for a variety of reasons. While some use it for its ability to personalize content and automate tedious tasks, others have been found using it to cheat in classes [16]. This has led to discussions among faculty and administrators about the ethical use of ChatGPT in educational settings [17].

While existing studies have explored the applications of ChatGPT in education, there is limited research on the motivations driving student usage. The Uses and Gratifications Theory (UGT) provides a framework for understanding these motivations but remains underutilized in this context. Students use ChatGPT for cognitive gratifications (information-seeking), affective gratifications (emotional satisfaction), personal integrative gratifications (self-awareness), and social integrative gratifications (peer interaction). However, the extent to which these motivations influence learning outcomes or ethical behavior remains unexplored. This study aims to explore the factors of using ChatGPT that facilitate academic assistance in higher education and to explore the ethical implications of ChatGPT usage in academic settings.

Literature review

The integration of AI tools into educational environments has significantly accelerated since the advent of advanced AI models like ChatGPT-4, released in March 2023. The use of AI tools such as ChatGPT and intelligent tutoring systems has revolutionized personalized learning by customizing educational content according to individual student needs. These tools provide real-time feedback, adaptive learning paths, and tailored resources, facilitating self-paced learning and addressing specific academic challenges effectively. For instance, OpenAI’s GPT-4o effectively guides students through complex problems without directly providing answers, thereby encouraging critical thinking and learner autonomy [20,21].

Beyond enhancing student learning experiences, AI technology has streamlined numerous administrative tasks. Educators increasingly leverage AI for automated grading, lesson planning, and assessment creation, freeing valuable time for interactive teaching and mentorship roles [22,23]. A notable development is ChatGPT Edu, launched in 2024, specifically designed to assist universities in grading and personalized tutoring, demonstrating AI’s growing role in educational management [20].

AI tools have notably expanded inclusivity in education, employing technologies such as text-to-speech, visual recognition, and adaptive interfaces, thereby addressing the needs of neurodiverse students and those with disabilities. These innovations help reduce educational disparities, ensuring broader access to quality education [24]. Moreover, AI-driven platforms such as Canva Magic Write and Curipod are facilitating student collaboration and creativity, assisting them in generating high-quality instructional materials and assignments efficiently [25].

Recent literature provides an interdisciplinary perspective on AI’s transformative impacts, advocating a human-centered approach for its integration into education [26]. Research highlights that ChatGPT’s context-aware responses significantly enhance student engagement and reduce instructional burdens, underlining the importance of balanced integration, digital literacy, and authentic assessment practices [27]. Meta-analyses further confirm that AI tools significantly boost cognitive, behavioral, and emotional engagement, demonstrating their role as tutors, assistants, and collaborators [28]. A synthesis of 21 case studies employing pedagogical models (TPACK, SAMR, Laurillard) further underscores how generative AI supports personalized feedback, collaborative writing, and critical thinking skills development [29]. All these connects ChatGPT use to emotional and behavioral engagement, suggesting deeper psychological and motivational dynamics behind usage patterns [28].

Recent survey data reflects a significant rise in AI adoption among students, with the Digital Education Council’s 2024 survey reporting that 86% of higher education students actively use AI, predominantly ChatGPT, for academic purposes such as summarizing texts, paraphrasing, and drafting essays. Despite this extensive adoption, approximately 58% of students still feel inadequately prepared for an AI-driven workforce, highlighting a critical gap in AI literacy and underscoring the urgent need for comprehensive AI training within educational curricula [30,31].

The ethical implications surrounding AI usage in academia are pronounced, particularly concerning academic integrity and plagiarism. A study by Welding revealed that while many students view AI tool use as cheating, a significant 43% still employ these tools for assignments and examinations [32]. Universities such as Flinders and Adelaide have adopted policies permitting regulated AI usage in academic work, conditional on adherence to clear guidelines, stressing critical thinking verification exercises alongside AI-generated outputs [33]. The University of Queensland advises students to verify their course requirements before incorporating AI in assessments to emphasize the importance of critical thinking [34].

Interactive textbooks and chatbot applications have become widespread, significantly contributing to student engagement by simulating human-like dialogues that promote language acquisition and communication skills [35–37]. Nonetheless, challenges remain, particularly regarding the optimal integration of chatbots into education without negatively impacting the learning experience [37,38].

Personalized learning facilitated by AI addresses individual educational requirements effectively, yet limitations such as the rigidity of AI-generated tasks and a scarcity of appropriate resources hinder full realization of its potential [39,40]. Educators integrating AI technologies into classroom management often encounter challenges due to limited understanding of these technologies, resulting in feelings of loss of control over teaching strategies [41–43].

AI-driven systems also benefit educators’ professional development through real-time data analysis, offering personalized feedback that improves teaching methodologies. Nevertheless, limitations remain concerning the breadth of AI-generated recommendations and the compatibility of predictive models with diverse student data sets [44–48]. AI-powered assessment tools significantly enhance grading accuracy and efficiency, though their application remains predominantly confined to specific areas such as language acquisition [49–51].

However, AI’s ease in content generation has amplified concerns regarding plagiarism, ethical misuse, and biased or inaccurate information outputs, necessitating robust guidelines and strategies for responsible AI integration into educational practice [30,52]. Studies from Southern Asian contexts indicate a correlation between AI usage and decreased academic integrity, driven by behavioral changes such as procrastination and diminished classroom engagement [53]. Experimental research further cautions that reliance on AI tools like ChatGPT can significantly diminish creativity and originality, especially within creative disciplines, underscoring the need for cautious integration and policy regulation [54].

To address these multifaceted challenges, institutions and policymakers must actively promote AI literacy and ethical standards. Initiatives like UNESCO’s AI competency framework and legislative efforts such as California’s inclusion of AI literacy in K-12 curricula underscore the importance of responsible AI use [20,54]. Furthermore, recent advancements in multimodal AI models (e.g., GPT-4o) present new educational opportunities through enriched audio, video, and image capabilities, necessitating careful implementation to ensure equity and accessibility [21].

While AI has substantially reshaped educational practices through enhanced personalization, administrative efficiency, and creative support, it concurrently poses significant ethical and practical challenges. Addressing these issues requires comprehensive AI literacy training, equitable access strategies, robust ethical frameworks, and ongoing research into AI’s long-term impacts on education.

Theoretical framework

The Uses and Gratification Theory (UGT) is a communication theory that was developed in the early 1940s by Katz and Blumler [55]. It seeks to explain why people use various forms of media and what gratifications they seek from those forms of media [56]. Following similar studies [57–60], this theory can be adapted to explore the motivations for using ChatGPT among university students.

Individuals actively choose and use various forms of media to satisfy their various needs and desires, as stated by UGT. The theory identifies four categories of gratifications that individuals seek from media: cognitive, affective, personal integrative, and social integrative [56]. Cognitive gratifications refer to the need for information, knowledge, and understanding [59]. Affective gratifications refer to the need for emotional experiences, such as pleasure, excitement, and relaxation [58]. Personal integrative gratifications refer to the need for self-awareness, self-esteem, and personal identity [57]. Social integrative gratifications refer to the need for social interaction, companionship, and a sense of belonging [60].

Motivation can be defined as the driving force behind an individual’s behavior or action [61]. In this study, motivation refers to the gratifications that students seek from ChatGPT, such as cognitive, affective, personal, and social integrative gratifications. Inspired by Urbancová and Fajčíková [62] A student may be motivated to use ChatGPT to seek information and knowledge on a particular topic (cognitive gratification), emotional experiences, such as humor or empathy, from the chatbot’s responses (affective gratification), self-awareness and personal identity through the chatbot’s responses (personal integrative gratification), or social interaction and companionship through the chatbot’s responses (social integrative gratification).

While UGT provides a robust framework for understanding media usage motivations, it has notable limitations when applied to the context of ChatGPT usage among university students. UGT primarily focuses on individual needs and gratifications, often neglecting broader social, cultural, and institutional factors that influence media consumption. For example, students’ use of ChatGPT may be shaped by academic pressures, institutional policies, or peer influence, which are not fully addressed by UGT. Although UGT explains why individuals use certain media, it struggles to predict how usage patterns evolve over time or in response to technological advancements. For instance, the rapid adoption of ChatGPT may be driven by trends rather than intrinsic motivations, which UGT does not adequately capture. UGT assumes that users are aware of their motivations, but unconscious or subconscious factors often play a role in media consumption.

Students may use ChatGPT impulsively or habitually without consciously seeking specific gratifications. The theory emphasizes the active role of users while overlooking the influence of media producers and the design of AI tools like ChatGPT. For example, ChatGPT’s user-friendly interface and its ability to generate instant responses may shape user behavior beyond their conscious choices. Research relying on self-reporting methods under UGT may lead to biased or incomplete data regarding motivations. Students might underreport unethical uses (e.g., plagiarism) or overstate productive uses due to social desirability bias [63].

In the context of this study, university students may use the ChatGPT to satisfy their cognitive gratifications by seeking information and knowledge on various topics. They may also use the ChatGPT to satisfy their affective gratifications by seeking emotional experiences, such as humor or empathy, from the chatbot’s responses. Moreover, students may use ChatGPT to satisfy their integrative gratifications by seeking self-awareness and personal identity through the chatbot’s responses. Finally, students may use ChatGPT to satisfy their social integrative gratifications by seeking social interaction and companionship through the chatbot’s responses.

The current study identifies four primary categories of gratifications: cognitive, affective, personal integrative, and social integrative. All of them derived from UGT and connects them to specific motivations for using ChatGPT.

For Cognitive Gratifications, students use ChatGPT primarily for seeking information and knowledge. For example, completing assignments by generating ideas or summarizing key points or preparing for exams by creating concise notes. UGT informs research questions about how students perceive ChatGPT as a tool for enhancing academic productivity.

For Affective Gratifications, emotional experiences such as humor or relaxation are sought by some students engaging with ChatGPT informally, such as asking humorous questions or seeking entertainment during leisure time or using ChatGPT responses to alleviate stress during tight deadlines UGT can help to explore how emotional satisfaction influences the frequency and type of ChatGPT usage.

For Personal Integrative Gratifications, some students seek self-awareness or personal identity enhancement through ChatGPT, like career planning advice and decision-making assistance (e.g., choosing TV shows or recipes). UGT helps frame questions about whether such usage fosters self-confidence or dependency on AI.

For Social Integrative Gratifications, social interaction and companionship are reflected in students’ discussions about ChatGPT with peers, such as sharing prompts and results within friend circles or recommending ChatGPT as a resource for academic tasks. UGT can help investigate how social dynamics shape attitudes toward AI tools. By integrating these dimensions into the study design, UGT enables a nuanced exploration of how diverse motivations drive the usage of AI tools like ChatGPT in educational settings.

1. Research Objectives and Questions:

As guided by the UGT and based on the literatures, this study attempts to explore university students’ perceptions and practices regarding the use of ChatGPT, identify the motivations behind its academic and non-academic applications, and examine the ethical implications and challenges related to academic integrity and responsible AI use. Therefore, the following research questions are proposed:

RQ1. How do university students perceive and practice the use of ChatGPT in academic and non-academic contexts?

RQ2. What ethical and pedagogical challenges emerge from its integration in higher education?

Method

To answers these research questions, this study adopted a qualitative approach, specifically, an exploratory study inspired by a Gruzd [60] to collect data from university students who used ChatGPT to understand their motivations. Qualitative research is well-suited for examining individuals’ subjective experiences, attitudes, and motivations, allowing for in-depth exploration and rich data collection [64]. In this study, the primary method utilized is in-depth interviews and semi-structured questionnaire, which provide a comprehensive understanding of the participants’ perspectives and motivations related to ChatGPT usage.

For this reason, purposive sampling was used in this research instead of random sampling to get our data. It is a non-probability sampling technique, to select participants who met specific criteria relevant to the research objectives. Purposive sampling allowed the researchers to deliberately recruit participants who were regular students at Khulna University and had used ChatGPT. This approach was both methodologically appropriate and logistically convenient, as both researchers are affiliated with the same university. By targeting students with direct experience using ChatGPT, the researchers could gather data that directly addressed the study’s UGT framework.

20 semi-structured in-depth interviews were conducted as part of the exploratory study with students at both the undergraduate and graduate levels. The size of the sample was determined based on data saturation, which is the point at which the interviews no longer provide any new insights or information. Previous researchers found that data saturation in qualitative interviews often occurs within the first 12 interviews, and basic themes can emerge as early as six interviews. Previous research demonstrates that “9–17 interviews” are typically sufficient to reach saturation, especially for studies with a homogenous population and narrowly defined objectives [53]. This ensured that the samples of the current study was adequate for capturing a comprehensive range of perspectives without unnecessary redundancy.

Due to the researchers’ convenience, the participants in this study consist of only Khulna University students. There were two inclusion criteria: (1) the Participant must be a regular student at Khulna University, and (2) must have experience using ChatGPT at least 3 times. The purposive sampling technique was employed to ensure diversity in terms of age, sex, academic disciplines, and levels of study. Each member was selected from different disciplines, including humanities, social sciences, natural sciences, and professional fields like engineering and architecture. This ensured that the findings reflected a wide range of academic contexts and needs. Also, participants included both undergraduate and graduate students, ensuring representation across different stages of academic progression. This allowed the study to capture variations in ChatGPT usage based on academic maturity and workload, see Table 1. The participants’ ages ranged from 21 to 28 years old. The breakdown for sex was 63.64 percent males and 36.36 percent females. Participants were chosen to reflect diverse usage scenarios of ChatGPT, such as academic tasks (e.g., assignments, thesis writing), personal development (e.g., career planning), and leisure activities (e.g., entertainment). This enriched the dataset by incorporating varied motivations and experiences.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Background Information of the Participants.

https://doi.org/10.1371/journal.pone.0333089.t001

Bangladesh presents a unique context for studying ChatGPT’s impact due to its evolving educational landscape and challenges such as large classroom setups, resource shortages, and low student participation rates [12]. There are several factors justify focusing on Bangladeshi students. Bangladesh is undergoing rapid technological transformation across various sectors, including education. University students are increasingly adopting AI tools like ChatGPT to address resource gaps and enhance academic productivity [13]. The socioeconomic conditions in Bangladesh make cost-effective tools like ChatGPT highly attractive for students who may lack access to premium educational resources such as paid writing assistants or tutors [12]. Bangladeshi students often face linguistic barriers when accessing global educational materials. ChatGPT’s ability to provide translations and summaries tailored to local contexts makes it particularly valuable in this setting [12]. Large class sizes and limited teacher-student interaction hinder personalized learning in Bangladeshi universities. ChatGPT offers solutions by providing immediate feedback and interactive support that can mitigate these issues [13].

Following an analysis of the pertinent literature, a semi-structured interview questionnaire was developed, see Supporting Information (S1 File). This questionnaire consisted of two modules that were incompatible with one another, the first of which was concerned with collecting primary data from the participants and was centered on demographic information. The second module consists of questions about the participant’s understanding, experience, and concerns about ChatGPT. It focused on students’ cognitive, affective, personal, and social needs, derived from the U&G theory within the educational context.

Results

The RQ1 asked how do university students perceive and practice the use of ChatGPT in academic and non-academic contexts? The interviewees answered the following about the following:

Discussion

This study reveal that students used ChatGPT for both academic and non-academic purposes (RQ1). Academically, it supported assignment writing, idea generation, presentation preparation, and even replaced traditional note-taking and writing tools like grammar checkers and paraphrasers. It also served as an alternative to search engines by offering direct answers without browsing multiple sources. Non-academic uses included writing job applications, emails, and social media content, as well as casual entertainment and decision-making support. However, ethical concerns emerged, as some students admitted to using ChatGPT for plagiarism in assignments and open-book exams, highlighting a blurred line between academic support and misconduct (RQ2).

The majority of the interviewees believed that ChatGPT is more efficient and effective than existing methods, this supports the study of Palasundram [36], Porter and Grippa [69], Niloy [70], and Belkina [29]. They have found that ChatGPT has been used to provide guidance and feedback. On the other hand, some of the students complained about how ChatGPT gives false references, made-up results require multiple prompts to get a near result, has login issues, and sometimes stops responding. Similarly, Holstein [71], Farrokhnia [27], and Vazquez-Cano [38] argued about the difficulties and limitations of human-AI conversation. They revealed that the feedback that was provided by these systems was pre-prepared and did not fulfill the requirements of every student. This study also found that several students admitted to using ChatGPT for assignments, presentations, and even open-book exams, raising serious concerns around plagiarism and academic dishonesty. Despite widespread perceptions of AI-assisted work as cheating, 43% of students still reported using such tools in academic tasks [32]. This highlights a growing tension between ethical standards and practical utility, as students prioritize efficiency over integrity.

Although Chew and Chua [37] found the time and reason to use chatbots to promote student learning and engagement unclearly, our study showed that it can be applied both in and off the classroom, even for daily life learning purposes. Students have used ChatGPT to build their daily schedules for studying or entertaining. Some previous studies [70,39,41,42,72–74] have shown that AI has been used to customize student learning assignments by teachers or university authorities. Recently, some universities permitted students to use ChatGPT with the declaration [33], also they need to provide references as well as put their argument and exercise critical thinking [34]. Some participants stated that they took the initiative and openly used ChatGPT during presentations and class tests. Though all the interviewees used ChatGPT to write assignments and some of them used it to conduct their research, they kept the information confidential as it would be considered plagiarism. This study found no integration of ChatGPT from Khulna University in terms of student usage.

The study highlights significant limitations of ChatGPT, such as generating false references, outdated information, and biased responses. These issues stem from the model’s design, which relies on probabilistic predictions rather than factual verification. The “hallucination” of fake references, for instance, is a byproduct of the model’s lossy compression and inability to verify the accuracy of its outputs [75,76]. This limitation raises concerns about the reliability of AI-generated content for academic purposes and aligns with broader debates in the literature about the risks of over-reliance on generative AI in education [10,28,77]. Additionally, biases in ChatGPT’s training data may perpetuate stereotypes or cultural prejudices, which could negatively impact equitable learning environments [27,78].

Conclusion

This study sheds light on the multifaceted motivations driving university students’ engagement with ChatGPT, highlighting both its academic and non-academic applications. Despite the tool’s perceived efficiency in enhancing productivity and facilitating various tasks, the findings emphasize critical ethical concerns, particularly relating to academic integrity and the potential weakening of critical thinking skills.

Notably, the study has significant limitations. Primarily, the sample size of 20 participants from a single university in Bangladesh limits the generalizability of the findings. Given the global and diverse nature of ChatGPT usage, the narrowly defined geographical and cultural context restricts the applicability of the results to broader, international student populations. Furthermore, the small sample heightens the possibility of skewed perspectives, influenced by local educational practices, technological accessibility, and socioeconomic conditions.

Moreover, this research captures self-reported perspectives and behavioral intentions, which might differ significantly from actual behaviors. Students’ claimed intentions and experiences with ChatGPT could potentially differ from observed usage patterns. Thus, future research should expand this qualitative exploration to larger, more diverse populations, using mixed-method approaches that incorporate observational or experimental methodologies to validate reported behaviors against actual usage.

Future studies should also consider longitudinal designs to track changes in student interactions with AI over time, especially in response to technological advancements and evolving educational policies. Given these limitations, future research should utilize this study primarily as a methodological starting point, emphasizing the importance of rigorous qualitative techniques to explore complex phenomena like AI adoption in education, rather than drawing definitive conclusions from the specific findings presented here.

Overall, this research advocates for a balanced and ethically-informed integration of ChatGPT in educational practices. By proactively addressing limitations, clearly defining usage guidelines, and focusing on AI literacy, educational institutions can effectively harness AI’s potential while safeguarding academic standards and fostering robust critical thinking skills among students.

Supporting information