Teachers’ and students’ use of ChatGPT at Social science faculty in the public and private Universities of Bangladesh

Introduction

ChatGPT, an AI technology bearding the GPT (Generative Pre-trained Transformer) architecture, can reasonably improve the teaching and learning process due to its sage conversational capabilities (Baidoo-Anu & Ansah, 2023). It can provide translation, summary, and question answers, and simultaneously, the generation of texts automates pestering (Cotton et al., 2023). Unsupervised pre-training and controlled fine-tuning generate responses similar to what a human expert would say, displaying domain knowledge about various topics and answering questions accordingly. Textual data is gathered from various internet sources, including websites, articles, papers, and forums (Dwivedi et al., 2023). The most remarkable feature of ChatGPT is its ability to generate text. Within just three months of the collective release of ChatGPT, significant numbers of software engineers, authors, academics, teachers, and songwriters used the system to generate written content, academic essays, computer programs, applications, and song lyrics (Mariani et al., 2023). ChatGPT-4 has continued to be developed, and today, it is used across different applications, including education worldwide. It also helps promoting students’ educational experiences through tailored coaching and support. Implementing this technology has advanced student engagement and collaboration by allowing students to question and debate asynchronously, removing the need to be simultaneously present (Li & Xing, 2021). ChatGPT can facilitate teachers and students discussing complex ideas at their preferred pace. It also helps with administrative tasks such as responding to commonly asked questions, providing course materials, and managing scheduling or registration tasks. Universities can utilize ChatGPT to improve student engagement by providing continuous support and accelerating administrative processes (Biswas, 2023).

However, despite these advantages, several critical concerns must be taken under consideration before utilizing ChatGPT. In the context of educational assessment, plagiarism comes as a threat. AI essay-writing systems produce essays based on present parameters or guiding principles which can result in academic forgery by submitting plagiarized write-ups. It counters the key central idea of education, which is to educate and inspire learners, and it could undermine the credibility of academic degrees in the coming years (Dehouche, 2021). So, the role of ChatGPT in education has sparked and initiated one of the most talked about debates in academia. Therefore, further research is needed to determine ChatGPT adoption in developed and developing countries. An in-depth analysis shows that comparatively, more affluent countries are moving quickly on the chatbots issue and developing ways to leverage it for education. However, developing countries such as Bangladesh need to focus on this. Because technologies do not have borders, understanding the utilization of ChatGPT by a diverse country like Bangladesh could provide a worldwide perspective of the benefits and challenges of adopting and adapting AI technologies in multiple socio-cultural and economic environments. Bangladesh is a developing country where the educational system is being modernized, and technology is being introduced in the classrooms of public and private universities (Sarker et al., 2019).

Recently, a growing number of teachers and students at tertiary education sector in Bangladesh have been shown interested in using ChatGPT. By focusing on this specific context, the study aims to bring insight into the perception and integration of emerging advanced technology such as ChatGPT into the educational practices of an emerging country. Focusing on Bangladesh as a specific research context, the study aspires to produce meaningful and enlightening knowledge for the worldwide community. In the backdrop of social science faculty at the public and private universities of Bangladesh, where teaching and personalized learning pathways are of great importance, it is essential to address how teachers and students are using ChatGPT. Comparatively, the educational system of Bangladesh has its drawbacks, with large classroom setups, shortage of resources, and potentially low student participation (Prodhan, 2016). These issues may be mitigated through ChatGPT, which provides personalized support, immediate feedback, and a highly interactive learning environment. Understanding the perception of the key stakeholders’ regarding the use of ChatGPT is crucial to harness this type of advanced technology effectively. Under this circumstance, getting insight into teachers’ and students’ perspectives on employing ChatGPT cannot be neglected because they are an indispensable part of the teaching-learning process.

There is a lack of knowledge about the perceptions and use of ChatGPT among teachers and students in the general settings of social science faculty at public and private universities in Bangladesh. In order to comprehensively evaluate the potential and challenges of implementing ChatGPT in the teaching-learning process in developing countries such as Bangladesh, it is crucial to gain insights into teachers’ and students’ perceptions and usage of ChatGPT. Understanding the perspectives and usage of ChatGPT by teachers and students at social science faculty in developing nations on ChatGPT is essential to ensure its implementation aligns with their specific needs and circumstances. Thus, the aim of this study is to investigate the perceptions and usage of ChatGPT among teachers and students of the social science faculty at public and private universities in Bangladesh, aiming to fill the research gap and contribute to policy-making at universities regarding the use of ChatGPT. Based on this aim, the study addresses the following research question: What are the perceptions and usage of ChatGPT among teachers in the social science faculties of public and private universities in Bangladesh?

Literature review

Researchers have underscored the significance of understanding the perspectives of stakeholders in order to successfully facilitate the adoption and implementation of AI tools in educational ecology. The application of Artificial Intelligence (AI) in the field of education is an area that elicits controversy and disagreement. According to Rawas (2024), AI systems like ChatGPT have the potential to endanger the privacy of stakeholders, including teachers and students. To demonstrate that ChatGPT is being used as intended and does not have any unintended negative impacts on students or teachers, its use must be transparent and accountable. Moreover, Samala et al. (2025) have mentioned that using ChatGPT raises problems due to specific unfavorable attributes, including technological constraints, ethical dilemmas, plagiarism, deceit, and abuse. In addition, Chassignol et al. (2018) think that these technologies can also adversely impact human interactions, relationships, and the quality of learning. Furthermore, it neglects social engagement and the communal elements of the learning process. Sok and Heng (2024) have raised concerns regarding security, privacy, trust, inaccurate information and an overreliance on technology.

Conversely, numerous studies have focused on the advantages of AI and the positive perceptions of AI held by instructors and pupils. Research conducted by Zawacki-Richter et al. (2019) indicates that teachers generally hold favorable views toward AI and consider it a helpful tool for enhancing educational practices and increasing student engagement. Kaswan et al. (2024) state that AI can offer adaptable feedback, personalized learning opportunities, and intelligent coaching methods. Therefore, students typically hold favorable views toward AI and perceive it as a valuable tool that enriches their educational experiences. They value the ability of AI to offer personalized learning suggestions, immediate feedback, and engaging learning settings. It enhances students’ self-assurance in their aptitude to acquire knowledge and renders the studying process more captivating for them. Vázquez-Parra et al. (2024) have found that university students positively perceived artificial intelligence (AI) as a valuable tool for learning. AI is capable of effectively addressing any inquiries related to topic comprehension.

Numerous AI-based technologies have been developed for educational purposes. Chatbots, an artificial intelligence (AI) tool, offer multiple benefits to education, assisting teachers and students in diverse ways. Sok and Heng (2023) indicate that ChatGPt helps with brainstorming, developing learning assessments, improving pedagogical practices, providing virtual personal tuition, and constructing an outline for an essay or research article. In addition, Chatbots provide tailored learning experiences, support students, improve tutoring and homework assistance, facilitate language acquisition, enable assessment and feedback, enhance student involvement, and aid in research and information retrieval (Pérez et al., 2020). ChatGPT is gaining popularity in educational environments due to its advanced artificial intelligence capabilities. The GPT series developed by OpenAI, including ChatGPT, utilizes robust and extensive language modeling approaches. As a result of being trained on a wide range of varied texts, the AI model can respond in a manner that resembles human-like behavior and demonstrates a heightened level of comprehension of language (Dentella et al., 2024).

ChatGPT stands out due to its capacity to enable more sophisticated and contextually relevant conversations. ChatGPT’s remarkable versatility enables it to actively participate in diverse educational endeavors such as tailored learning, student assistance, tutoring, and evaluation. Due to its capacity to comprehend intricate inquiries, deliver precise information, and provide comprehensive arguments, it enhances the efficacy of learning. Kasneci et al. (2023) have found that ChatGPT can aid educators and learners in improving their proficiency in language acquisition, data comprehension, analytical thinking, logical reasoning, and academic research. Although ChatGPT holds significant value in education, it has faced substantial criticism due to its potential to undermine academic integrity. Several issues that require attention include copyright infringement, discrimination, inequity, data privacy and security concerns, and excessive dependence of students or teachers on ChatGPT (Wu et al., 2024).

Shoufan (2023) conducted a study to determine students’ perceptions of using ChatGPT. The findings demonstrate that students are impressed by the capabilities of ChatGPT and its utility as a tool for studying and working. Users value the humanoid interface of the system for its user-friendly nature and ability to provide well-structured and logically sound responses. Nevertheless, many students contend that the responses provided by ChatGPT lack consistent accuracy. Technology cannot substitute human intelligence, and it is widely accepted that a solid knowledge base is crucial for effectively engaging with it. Firat (2023) suggests that incorporating ChatGPT into education presents numerous opportunities to enhance learning outcomes, tailor instruction, and transform the role of teachers. However, this progress needs to be improved regarding assessment, digital literacy, and ethical considerations. It is essential to maximize the capabilities of ChatGPT entirely by tackling these issues and devising strategies to ensure ethical and fair implementation.

According to a study conducted by Mogavi et al. (2023), the aspects of ChatGPT that are most commonly disputed are creativity and ethics. While ChatGPT is seen as a pioneering advancement by certain initial users, with the potential to boost students’ self-confidence and motivation to learn, there are apprehensions that excessive dependence on the AI system might promote superficial learning habits that could impede students’ critical thinking skills. In another study, Cotton et al. (2023) have presented a series of tactics that institutions can adopt to guarantee the ethical and conscientious utilization of ChatGPT. The strategies encompass implementing standards and processes, offering training and support, and employing various tools to identify and deter cheating.

To comprehensively evaluate the potential and challenges of implementing ChatGPT in developing countries such as Bangladesh, it is crucial to gain insights into the perceptions of teachers and students towards this technology. Gaining insight into the perspectives of teachers and students regarding the assistance provided by ChatGPT in developing nations necessitates a substantial investment of time. In financially constrained countries, access to resources and opportunities for obtaining high quality education is limited (Kshetri et al., 2023). Understanding the perspectives of teachers and students in developing nations on ChatGPT is essential to ensure its implementation aligns with their specific needs and circumstances. These findings inform the creation of effective integration strategies for settings with few resources, and ChatGPT has the potential to address educational gaps and offer customized learning opportunities. Nevertheless, the perspective of teachers and students at public and private universities in Bangladesh towards applying ChatGPT as an artificial intelligence tool for educational purposes remains largely unexplored.

To the best of the authors’ knowledge, more research is needed on ChatGPT in Bangladesh. Islam and Islam (2023) conducted a study to ascertain the advantages and disadvantages of implementing ChatGPT in educational settings in Bangladesh. To accomplish that goal, data were obtained from online sources and processed utilizing the noticing-collecting-thinking model. The outcomes of this AI strategy are divided into four categories: research, education, the development of individual skills, and social. Rahman and Watanobe (2023) conducted a study examining the potential benefits and risks that ChatGPT could have on education in Bangladesh. Furthermore, researchers investigated how ChatGPT improves students’ programming skills. They employed ChatGPT to conduct various coding-related experiments, such as generating algorithm pseudocode, rectifying code errors, and generating code based on problem descriptions. The generalizability of the study’s findings to teachers and students engaged in technology-related topics across different disciplines is limited.

Rahman et al. (2023b) carried out a study demonstrating the use of ChatGPT in academic research. They presented a practical example and offered suggestions. Published articles, websites, blogs, and visual and numerical artifacts were all used to gather the data for this study. The findings indicate that ChatGPT can serve as a valuable instrument for creating initial notions in scholarly studies. The literature synthesis, citations, problem descriptions, research gaps, and data analysis may present challenges for the researchers. These studies do not, however, focus on how teachers and students perceive the use of ChatGPT. Therefore, this study aims to ascertain how teachers and students at both public and private universities in Bangladesh perceive applying ChatGPT.

Theoretical framework

This study is grounded in the Unified Theory of Acceptance and Use of Technology (UTAUT) model, which posits four core constructs such as performance expectancy, effort expectancy, social influence, and facilitating conditions that are theorized to shape the acceptance and usage of technology (Venkatesh et al., 2003). Performance expectancy is defined as the extent to which ChatGPT is perceived to improve educational outcomes, such as improving teaching efficiency for teachers or helping students during the learning process. Effort expectancy answers how easily teachers and students can leverage ChatGPT effectively during their studies. The social influence reflects the effect of peers, faculty, and institutional norms on the decisions of individuals to adopt ChatGPT. Finally, facilitating conditions are the resources and support at the educational institution that enable the use of ChatGPT, such as training or technical support. Along with these constructs, the UTAUT model added two vital outcomes: behavioral intention and actual use (Rofiah & Suhermin, 2022). Teachers’ and students’ perceptions about ChatGPT and how they use ChatGPT were guided by behavioral intention and actual use outcomes to fulfill the objectives of this study ( Figure 1).

Figure 1. Theoretical framework of this study.

This figure delineates the theoretical framework that guides this study. This framework is based on the Unified Theory of Acceptance and Use of Technology (UTAUT) model proposed by Venkatesh et al. (2003).

Methods

This study followed a mixed method approach combining both quantitative and qualitative data to get a comprehensive understanding of teachers’ and students’ perception and usage of ChatGPT in social science faculty of both public and private unviersities in Bangladesh. This approach allow the researchers to generalize the perception and usage of ChatGPT by teachers and students along with the exploration of underlying views, experiences, and contextual factors of adapting ChatGPT. Heyvaert et al. (2013) argue that mixed-methods research is crucial because it combines quantitative data with qualitative data, which may provide a holistic understanding of complex phenomena. This strategy demands for a more in-depth exploration of findings. As a result, this study followed a mixed-method approach to generalize social science faculty teachers’ and students’ perceptions and use of ChatGPT in Bangladesh and also to comprehensively get an in-depth scenario of their experiences, strategies, and interactions. Two survey questionnaires were generated for both teachers and students. Each survey question contains twenty-six items and is prepared on a five-point Likert scale. In order to gain a comprehensive understanding of the perspectives of teachers and students from public and private universities in Bangladesh, interviews were carried out. As a theoretical framework, the UTAUT model helps to guide the preparation of questionnaires and the interpretation of the findings. However, no copyrighted survey items or scales from the proprietary UTAUT instrument were used in this study; rather, the model was adapted conceptually to develop data collection tools and also cited properly.

The survey items were developed based on the UTAUT constructs, including performance expectancy, effort expectancy, social influence, facilitating conditions, behavioral intention, and actual use. Items were generated through a review of prior UTAUT-based studies on technology adoption and refined to capture context-specific aspects of ChatGPT use in Bangladeshi higher education. The content validity of the two survey questionnaires was ensured through getting reviewed by three EdTech experts. As the questionnaires were in Bangla (the mother tongue of the participants), two language experts also validated the questionnaires. Based on their feedback, questionnaires were revised before piloting. Moreover, Cronbach’s alpha was utilized to assess the reliability of the instruments. Forty teachers and fifty-three students participated in the piloting of two surveys. While the students’ survey had a Cronbach’s alpha of 0.833, the teachers’ survey had a Cronbach’s alpha of 0.877. These results show that both questionnaires are highly reliable for gathering data.

Based on piloting feedback, a few items were reworded for clarity, redundant items were removed, and the overall structure of the two survey questionnaires was slightly modified to enhance the clarity of the tools. Pilot data were excluded from the final analysis to avoid bias in the main study. E-survey link was shared through institutional email lists, social media groups, and direct communication channels with teachers and students. A stratified sampling technique was used to collect survey data using Google Forms due to its wide accessibility, user-friendliness, and efficiency in organizing responses. The survey link was shared through institutional email lists, social media platforms, and direct communication with teachers and students. In terms of sample size, data were collected from the social science faculties of sixteen universities, comprising eight public and eight private universities. In total, survey data were collected from 402 teachers and 440 students across these social science faculties. However, after data cleaning, the dataset comprised 381 teacher records and 384 student records. Gender and other demographic factors were considered when collecting data, including data from 174 female teachers and 171 female students.

Then data was analyzed through IBM SPSSversion 27 (IBM Corp, 2021) (https://www.ibm.com/support/pages/downloading-ibm-spss-statistics-27). For researchers who seek an open-source alternative, can use, JASP for similar data analysis. JASP is an open-access software (available at https://jasp-stats.org/).

In terms of qualitative data, a convenient sampling technique was used for conducting in-depth interviews due to the accessibility and availability of teachers and students within the study timeframe. This convenience sampling allowed the inclusion of teachers and students from both public and private universities who were available and willing to provide detailed insights about their perception and use of ChatGPT. A total of 32 interviews were conducted, comprising 16 teachers and 16 students, with two teachers and two students from each university, ensuring equal gender representation at each university. The interviews were recorded with permission.

Research objectives were explained to the participants and commitment of conserving privacy of the participants in every aspects of the research were declared. Interviews were taken only those who willingly showed interest to participate in the study. Each interview lasted for 35 to 40 minutes. After collecting data, every recorded interview was transcribed verbatim. Next, all Bengali transcriptions were converted into English and was sent these to a Bengali and English language specialist to ensure the accuracy of the translation.

Researchers completed transcriptions after getting the translator’s input. The research employed the thematic qualitative data analysis technique to identify patterns and theme to understand the real situation deeply (Braun & Clarke, 2006). By the six stages of data analysis outlined by Braun and Clarke (2006), all researchers first became familiar with the data by reading it separately, and then each researcher created preliminary codes from it. At this point, they adopted the strategy of “reading, reading, and reading again” (Mertler, 2009, p. 141) to examine all narrative data. Researchers grouped related codes into new themes in the third step and then reviewed them again in the fourth. At the fifth step, several themes were identified, and mutually agreed-upon final themes were assigned names. A report was then generated. Since researchers transitioned from a factual understanding to a theoretical one, they employed an inductive approach to qualitative data analysis (Graneheim, Lindgren, & Lundman, 2017). Direct quotes from teachers and students were used to highlight significant issues in the Findings section, thereby maintaining transparency and credibility. This approach allows participants’ voices to be authentically represented while facilitating the thematic interpretations.

Findings

The following ( Table 1) demonstrates the demographic information of the teachers.

Regarding the demography of teachers, it was found that public universities represented the majority (53%) among 381 participants, while 47% were from private universities. In terms of gender, 54% were male and 46% were female. Academically, assistant professors led by 38.8%, followed by lecturers of 34.6%, while associate professors and full professors made up 16.3% and 10.2%, respectively. In terms of academic discipline, Economics had the highest representation at 26%, followed by Political Science at 21%. Moreover, Mass Communication and Journalism, and Public Administration each accounted for 14.7% of the sample, while Anthropology had the lowest representation at only 1%. The majorities, 79.3% of teachers, have a free version of ChatGPT, and 20.7% have a premium version.

The following table ( Table 2) demonstrates the demographic information of the students.

Survey data on student demographic showed that 55.5% of the student participants were from public universities and 44.5% were from private universities. The study includes the same gender distribution, with males constituting 55.5% and females 44.5%. Regarding the academic level, 53.6% of the respondents were graduate students, and 46.4% were post-graduate students. Mass Communication and Journalism was the most represented department in this study with 12.8% of students, followed by International Relationships (10.7%) and Public Administration (10.4%). Development Studies was the least represented department with1.6%. As for the usage of ChatGPT, 97.1% students utilized the free version, and merely 2.9% were premium users.

Teachers’ responses to the survey are presented in Table 3.

Teachers considered ChatGPT positively regarding its effective integration into enhancing teaching materials (M=3.51, SD=0.807), easy use (M=3.58, SD=0.809), and less effort required for incorporation into the teaching methods (M=3.48, SD=0.841). ChatGPT was frequently used to generate teaching materials (M=3.73, SD=1.089) and to check the accuracy of information (M=3.88, SD=0.931) but rated lower for helping with complex research (e.g., literature reviews and methods) (M=2.91, SD=0.858). The confidence level for using ChatGPT for research was relatively low (M=2.80, SD=0.955). These findings indicate that ChatGPT lacks reliability when applied in research purposes. Teachers also reported low levels of institutional support (M=2.49, SD=0.996) and policy guidance for ethical use (M=2.47, SD=1.017). Similarly, colleagues’ support (M=2.69, SD=0.929) and financial support for premium usage (M=2.42, SD=1.075) were presented as insufficient. These findings suggest that contextual and institutional support are essential for effective utilization of ChatGPT. Teachers intended to continue using ChatGPT (M=3.62, SD=1.023) and use it more (M=3.68, SD=0.967). They often thought about ethical risks (M=3.59, SD=0.944) and accuracy verification (M=3.88, SD=0.931), indicating a cautious optimism. They proactively sought training opportunities (M=3.81, SD=1.043) and were careful of excessive reliance on ChatGPT for creativity (M=3.82, SD=1.015). These findings indicate teachers’ concern about ethical risks and the overuse of ChatGPT.

Students’ response to the items is shown in Table 4.

ChatGPT led to increased efficiency for students and better comprehension of the course material as students reported. It was beneficial with more complex subjects (M=4.47, SD=0.625) and ideas for project generation (M=4.44, SD=0.769). Furthermore, ChatGPT quickly and accurately helped with tasks (M=4.47, SD=0.729). Students often used it to complete assignments (M=4.31, SD=0.591) and prepare presentations (M=4.07, SD=1.078). Moreover, students reported ease of use when integrating ChatGPT into their academic activities, highlighting the low effort to use ChatGPT effectively (M=4.15, SD=1.051). They deemed themselves reasonably competent in using ChatGPT (M=4.17, SD=0.745) and frequently sought advice and tutorials from ChatGPT (M=3.89, SD=0.890). These findings indicate that students see ChatGPT positively because of support it provides for their academic activities. Moreover, Peer support for the use of ChatGPT was widespread (M=4.18, SD=1.066), but institutional support was perceived as insufficient. Clear ethical use standards (M=2.14, SD=1.039) and IT assistance provided by the universities (M=2.53, SD=1.006) received noticeably low ratings. Although indicating a sustained uptake of ChatGPT in academic settings, students expressed moderate to high intentions for continuous use (M = 3.49, SD = 1.035) and more future utilization (M = 3.41, SD = 1.116), they were cautious about overreliance (M=3.16, SD=0.959) and still focused on creativity and academic integrity. These findings suggest that although the students lack awareness about the ethical use of ChatGPT, they remain cautious for being overly dependent on it.

Table 5 shows the descriptive statistics according to six constructs.

The study uncovered distinct patterns among the six variables (PE, EE, SI, FC, BI, AU) highlighting differences in perceptions between the students and the teachers of different public and private universities. In terms of performance expectancy (PE), students at public universities had the highest mean (M=4.36, SD=0.58) for the usefulness of the system for improving performance. Students enrolled in private universities showed a mean score of 4.20 (SD=0.62). Public and private university teachers scored significantly less than the overall score (public university teachers: M=3.72, SD=0.64; private university teachers: M=3.60, SD=0.68). These findings suggest that ChatGPT contribute more to increasing students’ performance, whereas teachers from both public and private university perceive the contribution of ChatGPT in their own performance as limited. Furthermore, the effort expectancy (EE) trends mirrored those for Performance Expectancy (PE). Public university students obtained the highest (M=4.28, SD=0.57), followed by private university students (M=4.15, SD=0.59). The average score of the public universities teachers EE was 3.68 (SD=0.65), while the lowest average score, 3.55 (SD=0.70), belongs to the private universities teachers. These findings suggest that students can use ChatGPT with less effort compared to the teachers in both public and private universities. On the other hand, students from a private university showed the maximum mean score (M=3.85, SD=0.64) with the most significant perception of societal pressure to use the system, while public university students had a mean score of 3.72 (SD=0.63). Teachers scored lower on this construct— public university teachers scoring an average of 3.25 (SD=0.60) and private university teachers scoring an average of 3.18 (SD=0.65). In terms of social pressure, teachers reported less compelled to use ChatGPT than students in both public and private universities.

For the facilitating conditions (FC), scores were relatively low across all groups, reflecting a consensus on a lack of support/resources. The students from public universities scored slightly higher (M=3.12, SD=0.75) than those from private university students (M=3.05, SD=0.72). Teachers showed the lowest ratings: teachers in public universities (M=2.95, SD=0.68) and private universities (M=2.90, SD=0.70). However, unlike other constructs, teachers had a higher behavioral intention (BI) than students. The highest mean score (M=4.10, SD=0.65) was identified for public university teachers, followed by private university teachers (M=4.00, SD=0.68). Still, public university students scored an average of 3.88 (SD=0.66), and private university students scored an average of 3.80 (SD=0.68). Lastly, regarding actual use (AU), having little variability, all groups were impressively consistent in how they scored. Public university teachers averaged 3.75 (SD=0.58), slightly ahead of private university teachers (M=3.70, SD=0.60). Students showed similar relative scores, with average scores of public university students of 3.68 (SD=0.63) and private university students of 3.65 (SD=0.64). In both public and private universities, teachers and students need institutional support to use ChatGPT and they have high behavioral intention to use it in future.

The one-way ANOVA indicates that the differences between groups are significant for the constructs of Performance Expectancy (PE), Effort Expectancy (EE), and Social Influence (SI) (p < .001 for each. PE between-group data showed substantial variation (Sum of Squares = 204.717, F(3, 761) = 234.430, p < .001), indicating substantial group-level differences. EE also showed significant differences (Sum of Squares = 182.140, F(3, 761) = 147.827, p < .001) and SI (Sum of Squares = 144.782, F (3, 761) = 111.399, p < .001). Behavioral Intention (BI) was significantly changed, but less strong effect (Sum of Squares = 7.806, F(3, 761) = 4.302, p = .005). However, no statistically significant differences were found for Facilitating Conditions (FC) (F(3, 761) = 1.613, p = .185) or Actual Use (AU) (F(3, 761) = 1.051, p = .369), indicating that attitudes for these categories were consistent across groups. The results deliver a comprehensive overview of multifaceted heterogeneity in users’ attitudes and intentions across multiple cohorts highlighting common challenges in favorable conditions and actual usage behavior, particularly in terms of perceived usefulness, ease of use, and social influence.

Tukey’s HSD post hoc analysis showed significant group-level differences across constructs. Performance Expectancy (PE) and Effort Expectancy (EE): Students in both public and private universities scored higher than teachers (p < .001), where the highest scorers were the students of public universities. These findings suggest that students perceive ChatGPT as valuable tool for increasing their academic performance and they acclaim its easy interface to use while teachers are more concern about ethical use of ChatGPT and inaccurate information provided by ChatGPT which impact on their usage of ChatGPT. This gap highlights the need for students to focus on the ethical use of ChatGPT, while teachers need training to use it effectively. Students outperformed teachers (p < .001) based on Social Influence (SI), whereas private university students performed the best. These findings suggest that students’ use of ChatGPT is influenced by their peer networks; however, without institutional guidance, it may lead to excessive and improper use of ChatGPT. In contrast, teachers may face weaker peer pressure to use ChatGPT. By Behavioral Intention (BI): the teachers outscored students, and public university teachers outscored private university teachers. However, the difference was not statistically significant. These findings underscore teachers’ willingness to utilize ChatGPT for teaching and research purposes in the future, provided their concerns regarding ethical use and potential inaccuracies are addressed. These findings indicate that social factors influence students’ use of ChatGPT, whereas teachers’ use of ChatGPT is driven by their behavioral intention. Peer-driven initiatives are necessary for teachers’ collective engagement, while a clear institutional policy is needed for students to use ChatGPT responsibly.

Findings from Qualitative analysis

Demographic profile of the participants

Table 7. Teachers profile.

ID	Name (Pseudonym)	Gender	Teaching experience (Years)	University type
T1	Rahul	Male	5	Public
T2	Shilpi	Female	12	Public
T3	Baizid	Male	14	Private
T4	Sharmin	Female	3	Private
T5	Sanjoy	Male	23	Public
T6	Fahmida	Female	10	Public
T7	Rahim	Male	7	Private
T8	Sumaiya	Female	9	Private
T9	Kamal	Male	13	Public
T10	Aleya	Female	20	Public
T11	Kuddus	Male	2	Private
T12	Sokina	Female	11	Private
T13	Jabber	Male	15	Public
T14	Jorina	Female	3	Public
T15	Sumon	Male	5	Private
T16	Kulsum	Female	9	Private

Table 8. Students profile.

ID	Name (Pseudonym)	Gender	Level of study	University type
S1	Morjina	Female	Masters	Private
S2	Babu	Male	Honors (2nd year)	Private
S3	Hymonti	Female	Honors (4th year)	Public
S4	Rabbi	Male	Honors (1st year)	Public
S5	Nahar	Female	Masters	Private
S6	Milon	Male	Honors (3rd year)	Public
S7	Baby	Female	Honors (2nd year)	Private
S8	Arafat	Male	Masters	Private
S9	Hena	Female	Honors (1st year)	Public
S10	Mukul	Male	Masters	Public
S11	Asma	Female	Honors (4th year)	Private
S12	Motin	Male	Masters	Private
S13	Tanni	Female	Honors (2nd year)	Public
S14	Shahid	Male	Honors (1st year)	Public
S15	Naju	Female	Masters	Private
S16	Mollik	Male	Honors (3rd year)	Private

The following figure ( Figure 2) demonstrates a comprehensive framework developed from thematic analysis of qualitative data.

Figure 2. Comprehensive framework from thematic analysis.

This figure illustrates a comprehensive framework of this study, which emerged from the thematic analysis of qualitative data from 16 teachers and 16 students in in-depth interviews. It highlights seven key themes, including knowledge, attitude, and skills; influence on performance; effort to use ChatGPT; support and Accessibility; Social Influence; concerns for using ChatGPT; and challenges for intention and use of ChatGPT.

Discussion

Bangladesh, as a developing nation, is undergoing through a transformation since it has been trying to integrate technology into several sectors, including education. The swift progress in Natural Language Processing technology has allowed educational institutions to consider incorporating innovative ways of instruction and learning. Teachers and students in the tertiary educational institutions at Bangladoesh use ChatGPT for their academic needs. According to this study, students of both public and private universities use ChatGPT as a highly immersive platform, demonstrating a high-performance expectation across the platform. This finding aligns with the research done by Limna et al. (2023), which found that students think ChatGPT helps them learn complex subjects. However, performance expectancy scores were lower for teachers than students at both public and private universities. This discrepancy can result from teachers’ cautious views regarding ChatGPT’s effects on learning environments. According to study by Kasneci et al. (2023), teachers are more suspicious about the practical usage of ChatGPT in teaching and research, even when students adapt it to some degree due to possible benefits. According to the effort expectancy results for both student groups of this study, ChatGPT was user-friendly. This is in line with Ollivier et al. (2023), who highlight how easy it may be for students to use AI tools. Both public and private university teachers’ have moderate level effort expectancy scores that also reflected in Sok and Heng (2023) study while they resonates with the need for professional development that can boost confidence and ease of use of AI.

Social influence is an important factor for ChatGPT usage. The findings show a moderate social influence on students’ use of ChatGPT, indicating that peer support plays a significant role in students’ engagement with AI tools. These findings support earlier research suggesting that classroom social dynamics significantly influence students’ use of technology (Winstone & Boud, 2022). On the other hand, both teacher groups scored lower on social influence, indicating that their institutions and peers don’t encourage them to use ChatGPT. Dawa et al. (2023) emphasize that positive feedback from colleagues increases teachers’ ChatGPT usage and encourages them to adopt it. ChatGPT usage also depends on some facilitating conditions. All teachers and student groups have identified the availability of technology resources and the institution’s support for using ChatGPT in an educational setting as barriers and poor facilitating conditions. This is consistent with studies by Limna et al. (2023) that highlight the need to have sufficient infrastructure and support mechanisms to ensure successful technology integration. These results also highlight a critical area for educational institutions to address since the potential benefits of AI technologies for teaching and learning might be seriously hampered by a lack of supportive environments.

Teachers exhibited a stronger behavioral intention to use ChatGPT than their students, although students in both groups expressed a modest intention to use ChatGPT in the future. This means teachers are more committed to employing these tools in their practices, even when students desire to use them. This result is consistent with studies that indicate teachers are beginning to recognize the benefits of using AI tools to improve their teaching (Ollivier et al., 2023). However, it also emphasizes how crucial ongoing assistance and training are to guarantee this commitment. According to the actual use scores of this study, both groups of students frequently utilize ChatGPT for various academic assignments, which highlight how actively they engage with AI technologies in their studies. This is also consistent with research by Limna et al. (2023) that shows students want to use technology to improve their educational experiences. Though their ratings were lower than those of students, teacher usage rates were also quite high, suggesting impediments to integration into teaching roles that are in line with Sok and Heng’s (2023) findings.

In addition to the various benefits of employing ChatGPT for academic purposes, a few teachers agree that the utilization of ChatGPT carries specific adverse ramifications for both teachers and students. Excessive comfort in working environments may hinder pupils’ intellectual maturation by suppressing their need to engage in critical thinking. Moreover, Bai et al. (2023) found that an overabundance of ChatGPT usage leads to a diminished ability to engage in critical thinking. ChatGPT has been identified as the cause of decreased memory recall. Furthermore, it might also hinder the professional growth of educators. Teachers at private universities utilize ChatGPT to enhance their lecture preparation and expand their expertise in various subjects. Conversely, teachers at public universities believe that ChatGPT’s knowledge base is limited and primarily receptive rather than comprehensive. Garg et al. (2023) and Spennemann (2023) claim that though ChatGPT is receptive to information, its capacity to provide extensive knowledge on certain subjects is very constrained.

According to Castillo et al. (2023), most students utilize ChatGPT due to its rapidity and precision, which benefit their learning process. Being a user from Bangladesh, individuals occasionally need some help due to the limitations of ChatGPT in providing proper responses in the Bangla language. It frequently misinterprets orders given in Bangla, which aligns with the research conducted by Kolar and Kumar (2023) on Hindi, Telugu, and Kannada. Furthermore, the requirement of a ChatGPT Plus subscription presents an additional obstacle to utilizing ChatGPT. Bangladesh is classified as a lower-middle-income country. In Bangladesh, there are lots of resource constraints to use online for teaching-learning like network issues, internet connectivity and speed, financial barriers (Rahman et al., 2023a). Hernandez (2019) posits that the financial constraints faced by hinder its ability to adapt technology in several sectors where it is required effectively. Similarly, most teachers and students in Bangladesh need help to afford the expensive fees associated with premium usage, primarily due to budgetary constraints. Consequently, consumers encounter numerous problems when using the free edition of ChatGPT.

ChatGPT has been found to exhibit inaccuracies or misinformation. Since it can autonomously provide information that could be more reliable, furthermore, it occasionally shows a deficiency in comprehending the context of a query, leading to the provision of irrelevant responses. Moreover, it generates content that is speculative or fictional. Nevertheless, despite all the constraints, the students use ChatGPT without considering its drawbacks. As students get more familiar with digital tools, they may need to pay more attention to the increasing prevalence of faults in academic procedures. However, teachers have limitations in assessing the authenticity of their students’ assignments, reports, and monographs because of the absence of plagiarism detection systems and their limited accessibility. Consequently, adhering to institutional regulations and guidelines is necessary to utilize ChatGPT. Limna et al. (2023) have highlighted the necessity of establishing an institutional policy to mitigate the disadvantages associated with using ChatGPT.

Teachers possess knowledge of the potential limitations of ChatGPT when used in educational settings. Despite their continual usage, they have concerns that it may hinder their capacity to learn autonomously and resolve issues. Contemporary educators utilize ChatGPT to confine their expertise to a particular domain, unlike their predecessors, who would amass their information by perusing several books, publications, and articles. The overreliance on ChatGPT may undermine teachers’ capacity for critical thinking and independent inquiry. Teachers express concerns about their students’ overreliance on technology and artificial intelligence. Consequently, students’ cognitive skills, such as critical thinking and problem-solving, may be impaired. Nevertheless, there is a consensus among both educators and learners that educational establishments can mitigate the adverse consequences by fostering critical thinking, digital literacy, well-balanced incorporation of AI technology, and ethical considerations. Explicit policies must be formulated and executed to utilize ChatGPT in educational and research endeavors.

Conclusion

This study endeavors to investigate the perception and use of ChatGPT by social science faculty teachers and students at public and private universities in Bangladesh for instructional purposes. This study reveals several noteworthy concerns that require further contemplation and examination. The study findings suggest that students of both public and private universities have higher performance expectancy and effort expectancy compared to their teachers. Teachers hold divergent perspectives on using ChatGPT, with some displaying enthusiasm while others expressing skepticism. Nevertheless, students at both types of universities view ChatGPT favorably for a range of educational endeavors and learning prospects, such as improved knowledge and research capabilities. Students from both private and public universities affirm that utilizing ChatGPT improves their learning outcomes and positively impacts their academic achievement. ChatGPT is also helpful for teachers. It enhances their teaching efficacy by assisting in class planning and providing comprehensive expertise on the subject matter. In addition, students of both public and private universities are more socially influenced by their peer networks, while teachers face weaker peer pressure but have higher behavioral intention to use ChatGPT. Despite these differences, neither teachers nor students are satisfied with the facilitating conditions of using ChatGPT, such as institutional support. For example, due to the exorbitant cost of ChatGPT Premium, users often resort to using the free version, which offers limited functionalities. Robust networks and high-speed internet are essential for enhancing the accessibility of ChatGPT.

Furthermore, the efficiency of the English language and the limitations of other languages pose a significant obstacle to its effective utilization. This study also investigates the apprehension regarding the dependability of information offered by ChatGPT due to potential misinterpretation of queries. In addition, teachers express concerns over the ethical implications of utilizing ChatGPT and the potential for excessive reliance on it, students from both public and private universities still need to be made aware of these difficulties. Teachers concern about ethical use and over-reliance on ChatGPT reflect global debate on ChatGPT impact on academic integrity, assessment authenticity, and critical thinking ability (Kasenci et al., 2023; Rudolph et al., 2023). Besides, students limited awareness about ethical and responsible use of ChatGPT highlights urgent institutional policy to use ChatGPT. This study findings prioritize to eliminate barriers, including addressing ethical concerns, to ensure equitable access to ChatGPT and harness its potential to enhance the quality of teaching, learning, and research in higher education. The findings of this study will assist in other developing and underdeveloped countries with socio-economic and cultural contexts similar to Bangladesh’s, contributing to their higher education systems. In addition, this study will contribute in global literature on GenAI ethics in education. However, this study only focused on the social science faculty, teachers, and students of multiple universities in Bangladesh, so the findings may not reflect the holistic perspective of ChatGPT usage among all demographic user categories. Focusing on a single set of users limits the generalizability of the findings to a larger population because users from other areas of interest or geography may have completely different experiences and perceptions. Future studies could address these challenges and investigate ethical awareness, policy implication, and leadership role to proper use of ChatGPT.

Ethical approval statement

This study adhered to the ethical guidelines in the Declaration of Helsinki and received approval from the Institute of Education and Research Human Research Ethics Committee (IERHREC) with approval number 2023/01 on 10/02/2023. The protocol for the study was reviewed to ensure its adherence to ethical norms governing research involving humans. This review encompassed the protocol’s aims, methodology, and informed consent processes. All individuals who took part in this study provided their consent before participating, and their confidentiality and privacy rights were protected during every stage of the research procedure. All potential risks to the participants were minimized.

Informed consent

Researchers informed the study objectives and privacy conservation issue to the participants and provided written consent letter to the participants. After getting consent from the participants, researchers collected data.

Authors contribution statement

Conceptualization and planning by 1^st author, tools development by 2^nd author, data collection and data analysis by 3^rd and 4^th author; writing introduction, literature review, and implication by 1^st author, theoretical framework and methodology by 2^nd author, discussion by 3^rd author, conclusion by 4^th author; supervision, review and editing by 1^st and 2^nd author. All authors read the manuscript thoroughly and agreed to submit and publish in this journal.