Insights of higher-level educators on integrating AI technology into their teaching practices

Artificial Intelligence, AI, has been defined as the use of computer power to resolve complex decision making based on human intelligence (Berente, Gu, Recker, & Santhanam, 2021). AI uses software and hardware to create intelligent agents capable of vast and complex actions (Das et al., 2015). ChatGPT is an AI chatbot that was launched in November 2022. Chatbots usually have predefined answers which are used to respond to user inquiries. However, ChatGPT generates new content to interact with its users, using advanced AI programs called Large Language Models (LLMs) (Menon & Shilpa, 2023). It has become a disruptive technology globally, as it has changed the current mode of operation with a more superior alternative and now holds huge potential (Fernández-Batanero, Fernández-Cerero, López-Meneses, & Montenegro-Rueda, 2023).

Gao, Li, Shum, and Zhou (2020) noted that chatbots offer users flexibility and interactivity and thus provides a unique attribute to attract new users. Amidst its growing popularity, there are concerns surrounding its usage in education. Adepoju and Adeshola (2023) noted that academics were particularly concerned about assessment given the ease with which ChatGPT could solve assignment questions. However, Annamalai et al. (2023) postulated that ChatGPT could improve student engagement as it offers opportunities for personalisation. One can thus infer that ChatGPT has opened a new realm of possibilities in academia.

There is a dearth of information on the effective integration of AI tools as leaders and faculty do not have the benefit of learning from the experience of others when developing curriculum and pedagogy. Educators interacting with students would also add another dimension as it relates to the impact and efficacy of ChatGPT in the classroom as they get valuable insights. There was a need to understand how easy they think a tool such as ChatGPT can be harnessed to achieve teaching and learning objectives. Seeking input from educators on whether the popularity of this tool would affect their willingness also provided valuable insight as institutions consider ways to encourage adoption. Therefore, this study bridged the gap and added valuable insight on the willingness to adopt an AI tool for teaching and learning at tertiary level in Trinidad and Tobago.

It would be beneficial to acquire data to ascertain the benefits and application of a tool such as ChatGPT as an approved means of sourcing information for both educators and students. The focus of this study was on the educators' perceptions of how easy it is to use, and how willing they are influenced by those around to incorporate AI as a means of enhancing the teaching and learning experience. According to the UNDP (2024), AI technology integration in education, is needed to ensure that AI does not widen the gap between those who use it and those who don't, and also ensure risks are minimised due to possible misuse. The findings from this study can thus provide valuable insight to leaders in the Caribbean, to inform the creation and implementation of training programs, policy and procedures. This paper presented the results of the study concentrating on effort expectancy and social influence, key constructs of the Unified Theory of Acceptance and Use of Technology (UTAUT) model. Davis et al. (2003) developed the UTAUT model defining effort expectancy as related to the ease of use and ease of learning the system and social influence was related to the importance placed on a new system such the behaviour towards it was adjusted.

The objectives of the research were as follows,

1. To examine the ease of use and accessibility of ChatGPT as perceived by educators (Effort Expectancy).

2. To investigate the influence of social factors, such as colleagues' opinions and institutional support, on educators' decisions to use ChatGPT (Social Influence).

The investigation gathered information from educators at The NexTech University. The following questions were developed to guide this qualitative case study.

1. What are educators' experiences with the ease of use and accessibility of ChatGPT in their teaching practices? (Effort Expectancy)

2. How do social factors, such as colleagues' opinions, institutional culture and social trends influence college educators' decisions to use ChatGPT in their teaching? (Social Influence)

This literature review explores key studies and findings related to the effort expectancy and social influence constructs of the UTAUT model. It examines how they contribute to technology adoption across various contexts, focusing on their relevance in educational environments. The UTAUT model, shown in Figure 1, was developed by combining eight previously developed versions to address technology adoption (Sharawy, 2023). It provides a comprehensive framework for understanding the factors that influence technology adoption. Two critical constructs within this model – —effort expectancy and social influence – play a pivotal role in shaping individuals' decisions to embrace new technologies. Effort expectancy relates to the perceived ease of learning and using a technology, emphasizing how simplicity and user-friendliness can drive acceptance. Social influence, on the other hand, reflects the extent to which individuals perceive that other persons, such as peers, supervisors or societal norms, believe they should use a particular technology (Davis, Davis, Morris, & Venkatesh, 2003).

The strengths of this model include stronger predictive power for usage behaviours, applicability and trustworthiness (Momani, 2020). In contrast to these findings, Davis et al. (2003) have noted that there is complexity in the model, given that technology acceptance is moderated by age, gender, experience and voluntariness of the user. Although the UTAUT model has been lauded as an improvement in the TAM model, the simplicity of the framework in terms of logical decision making has also been noted as a limitation, since variables such attitudes and emotions are factors which should be considered when contemplating technology use, but are absent in the TAM based models (Bagozzi, 2007). Nevertheless, the model is still a useful tool providing a robust framework to assess technology acceptance and guide the extraction of data from educators using the key focus areas to create a rich account of their viewpoints on the use of this tool within their personal context.

According to the definition proposed by Sharawy (2023), AI is intelligence shown by machines instead of human beings. The option to embrace the AI technology in academia through integration, provides an opportunity to educators to become more productive and thus improve their quality of delivery (Chan & Tsi, 2023). Agrawal, Ali, Gans, and Goldfarb (2017) described the potential of AI technology to provide for a faster and cheaper alternative to derive content and course materials thus aiding in automating tasks. AI technology has proven to be beneficial through its predictive qualities using information that is currently available to generate new information (Agrawal et al., 2017). Whilst there may be a variety of viewpoints on the benefits of using AI which have been exposed to the public through research or other promotional methods, there is ambiguity as to the level of enthusiasm of educators in the local context to incorporate a generative AI tool for teaching and learning. Thus, there is a notable gap in the literature surrounding the local educators and their understanding of how to use AI technology to realise the benefits.

AI technology is evolving as new and improved options become available. One such example is a chatbot, an AI technology tool that can simulate a conversation with humans (Adamopoulou & Moussiades, 2020). It was built by using Natural Language Processing (NLP) to comprehend human language in order to produce a response. Das et al. (2015) indicated that AI chatbots are able to respond to topics based on the information it was trained on. This is a limiting factor in its usage as the tool refers to this information store, to build a response and interact with the user. Despite this limitation, the tool is an innovative one, as it provides new and unique opportunities for use in both the corporate and academic environments.

Given that chatbots are only as effective as the data they were trained on, the relevance of the response could be affected depending on the design. This could effectively hamper it's applicability in the classroom, if the output is outdated or contextually inappropriate. It could also affect the likelihood of educators to adopt it, if there is doubt as to the accuracy of information being provided. However, further research is necessary, to understand how the opportunities for improvement, can be harnessed and also the disposition of tertiary level educators to adopt such new technology.

Research conducted by Menon and Shilpa (2023) noted that the user-friendliness of the system could increase the probability of adoption. Effort expectancy, as the name suggests, refers to the ease of use of the system and this domain of the UTAUT model tests whether the system is clear and understandable, flexible and easy to learn. The research (Menon & Shilpa, 2023) has shown that effort expectancy does impact the use behaviour and thus acceptance of technology tools.

ChatGPT has a user-friendly interface which is easy to access via the internet and thus provides an excellent end user experience (Klimova, Pikhart, Shaikh, & Yayilgan, 2023). ChatGPT can also be used easily to analyse large datasets and create realistic scenarios as part of the learning curriculum in an automated way and brainstorm new ideas in project-based learning (Alshater, 2022). Al-Eyadhy, Alhasan, Altamimi, Jamal, and Temsah (2023) postulated that ChatGPT could support student learning as it provided a mechanism for accessing huge repositories of information while Atlas (2023) noted that the tool can save time and could potentially improve the quality of work. The UTAUT framework posits that minimal effort expectancy boosts users' willingness to accept and thus adopt technology tools.

The ease of use of ChatGPT to interrogate a vast data store to refine and produce content is appealing. However, this requires strategic integration and utilisation of the chatbot by educators, as it is not a seemingly obvious use of this AI technology tool. Even though there is a perceived ease of use of ChatGPT, the diversity in society today, means that there could still be varying levels of comfort by educators to integrate this new technology. Differing locales can have vastly differing perceptions on ease of use, even when using the same tool, hence the reason a study like this is quite timely.

According to Menon and Shilpa (2023), people are more inclined to use ChatGPT based on both social media (external) and peer influence (internal). Social impact may be affected by internal factors such as co-workers who have already adopted and are using the system, as well as leaders within the organisation. In the context of the UTAUT model, social influence refers to the degree to which persons are influenced by others who are deemed important to them, to adopt a new technology system (Davis et al., 2003). When senior colleagues are advocating the use of the tool then the social impact would be high (Davis et al., 2003). Therefore, the technology becomes more attractive because others are using it. ChatGPT has gained widespread popularity as demonstrated by the dramatic rise in the number of users since its launch.

This study employed a qualitative research design, utilizing a single case study approach to explore educators' perspectives on the adoption of ChatGPT for teaching and learning. The case study methodology is well-suited for gaining an in-depth understanding of complex phenomena within real-life contexts as it focuses on a holistic description and explanation. Heale and Twycross (2018) posited that the case study provides a structure for evaluation and analysis of the situation and the associated issues. This approach was especially useful in this educational research, where the context and the interactions were essential to understand the phenomena under study.

This qualitative research adopted the framework developed by Merriam (1988) and utilised purposive sampling for data gathering. It was necessary to select persons who were knowledgeable and willing to share their views on the topic at hand (Merriam, 1988). This study utilised non-probabilistic sampling as it was more suited to the qualitative methodology (Sheppard, 2020). The snowball sampling method was chosen as a way to extend the same population of persons who were knowledgeable and could provide valuable insight. Twelve participants were interviewed.

Semi-structured interviews were conducted virtually, allowing for continuous learning as the interviews progressed (Denzin & Lincoln, 2018) and exploration of topics raised. Participants were given pseudonyms P1 and P2 incrementally until P12, for the twelfth person. According to Mishra (2021), the data analysis framework developed by Merriam and Tisdell (2009) advocates for the researcher to simultaneously collect, code and analyse the data. The process of data analysis according to Merriam and Tisdell (2009) culminated with the creation of categories for the codes that were generated, followed by sorting of these categories into themes until a manageable level was reached and finally reduction of the themes.

In order to ensure quality of data and establish trustworthiness, Merriam and Tisdell's framework for qualitative research centres upon the ideas of credibility, consistency and transferability. Trustworthiness can be established by precisely following the defined process of collecting, ordering and disclosing the methods of analysis (Moules, Norris, Nowell, & White, 2017). Data source triangulation was used to ensure credibility and trustworthiness of the data by using two or more sources to corroborate the findings (Daniel, 2019). Transferability was determined in the richness of the data set and the resultant report of the analysis which allowed readers to apply the findings to their situation. Reflexivity was established through the use of a reflective journal to record the research journey including the decisions made, the rationale and the insights gained throughout the process.

For the purposes of this study, the researcher assumed that participants will be able were honest and responded objectively to the questions posed. It was also assumed that that the use of ChatGPT was constant during the data-collection period with no significant changes to the features and functionality of ChatGPT. This study was limited due to the sample size and generalizability. Give that the snowball sampling technique was used to select participants, there could be inherent bias in this technique as persons can choose others who can potentially skew the results. There was a potential for bias in responses, given the new technology being investigated, participants could have possibly provided responses that were more socially acceptable instead of their honest opinion. This is more commonly known as the “Hawthorne Effect” and was coined to describe a change in behaviour due to an acute awareness of being observed (Bendix & Wickström, 2000). The study was delimited by the qualitative methodology that selected to gather the experiences and perceptions of educators but not to provide statistical generalisations.

The participants had teaching experience ranging from 0 to 30 years, with four participants teaching for 1–10 years, six between 10 and 20 years and 2 persons having over 20 years as an educator. There was a mix of participants in terms of their background as just over 50% of respondents were based in technology discipline, whilst the rest were not in the technology discipline. They held roles such as teaching assistant, junior research fellow, instructor, lecturer and senior lecturer or professor at the NexTech University. The duration of the interviews was approximately 30–45 minutes. The secondary source of data that was identified was found on NexTech's website in the form of teaching tips and guidelines. It was found that there are no formal policies and procedures which dealt with the integration of technology tools such as AI in the curriculum.

The study interrogated participants on the effort required to use the tool. There was one emergent sub-theme which arose in this study, and many participants also shared their guidelines for using the tool. Figure 2 graphically presents an excerpt of the theoretical framework for effort expectancy.

Overwhelmingly, most of the lecturers and instructors within the technology discipline concluded that the tool was very easy to access and use. They noted that it was “pretty simple and straight forward”. Through a bit of practice, many concluded they were able to craft clever prompts to yield purposeful answers. P1 stated that ChatGPT was “much more tolerant of mistakes and it can infer what you are asking without being precise”. She noted that every iteration of the tool has resulted in improvements in terms of ease of use and user-friendliness. P5 and P9 liked that ChatGPT could be used in many ways and associated it to a “one stop resource” similar to the findings of Alshater (2022), who found that ChatGPT was an excellent resource as it could be used to automate tasks by creating content easily.

P2 remarked that the tool was quite novel, in that it “interacted with users as though it understands what you are searching for”. P4 remarked that ChatGPT was “a bit more dynamic, because we could interact with it” as he compared it with other teaching tools which are a bit more static in terms of the offering. The example he provided, was that through “prompt engineering we can do the same thing, but with a different scenario and get different output” (P4, participant interview May 24, 2024). Prompt engineering is a methodical approach for creating prompts to yield meaningful results from AI applications (Fidan, Işin, Işin, Işin, & Işin, 2024). P4 likened this feature of continuously interacting with ChatGPT to “spit balling” where you are conversing with the tool and throwing ideas back and forth to solve a problem. His enthusiasm was evident as he felt that this application of conversing with ChatGPT was “more authentic, because that is what you would do in the workplace” (P4, participant interview May 24, 2024).

P1 and P4, noted that ChatGPT was useful and easy to use, and gave targeted answers. Given their technology background and their understanding of how LLMs work, they were acutely aware of the best way to use the tool, aligned to its design and expected outcomes. However, it was not specifically created for use in academia. P6 noted that ChatGPT was very different from the game-based tools that she is accustomed to using. These game-based types of tools in her opinion, facilitated the three types of learning interactions (Learner/Learner, Learner/Content, Learner/Teacher) whilst ChatGPT did not. She did not believe that ChatGPT allowed for as much participation as the games-based tools she is accustomed to using.

While P9, P10, P11 and P12 also agreed that the tool was very easy to use, they too, struggled to assimilate the usefulness of the tool for academic related activities, as they thought the tool seemed to be more general. P11 admitted, that while he could use Prezi to create presentations for delivery of content in the classroom and Mentimeter to conduct polls to gauge student understanding, he had to think “outside of the box” to apply ChatGPT in the classroom. P11 concluded that ChatGPT would be a complementary tool to use for teaching and learning with these other tools. P12 also stated similar concerns when he revealed that other technology tools were very specific, for example, Learning Management Systems for managing courses and course content and the Zoom application to conduct online or virtual classes.

P11 revealed that he needed to evolve his assessments to more authentic assessments and having a tool like ChatGPT could certainly aid in making this task, less onerous. Whilst P11 recognised the need to adapt his assessments, he admitted that through reflection, he could learn how to use AI technology more optimally to produce the content required. Even as P11 conveyed his belief that modification of assignments was required to effectively use ChatGPT, P4 disclosed his ingenious approach. P4 was very excited to share his view of “embedding text” in his assignments as a way to “confuse ChatGPT”. He likened it to “poisoning the seed”, where the students who input assignments directly into ChatGPT (the seed) to generate answers, would not get an accurate response. He cleverly articulated his thought process to embed text by adding cultural words, or local terms into the assignments to cause ChatGPT to focus on those words, rather than the actual assignment. As he relayed his experience when testing this approach, P4 noted that ChatGPT returned an unsuitable response, that would not satisfy the requirements of the assignment. This therefore meant that students would not simply be able to input the assignment into ChatGPT and wait for the answer to be output, so that they could submit as their own work. In his mind, adopting an approach such as this one, would force students to critically analyse the assignment.

When asked to describe any challenges with the tool, P7 spoke about the repetitive content that she received, and also felt the content was outdated as it was trained only on data up until 2021. P1 highlighted her experience of using the tool and receiving content which was wrong. P1 understood however, that this was simply due to the data it was trained on, corroborating the view of Das et al. (2015), that this was indeed a limiting factor of the tool. P5 also echoed similar sentiments, when he stated that “the realm that we are in with ChatGPT, does not always given you the right answers…it's a large language model. It only knows as much as it is trained on”, (P5, participant interview, May 24, 2024).

Overall whilst the tool in isolation was simple and intuitive, participants revealed their challenges were associated with the application of the tool for teaching and learning. There were some examples shared by participants who were developing new strategies to incorporate the tool into their teaching “toolkit”. P11 revealed that he understood that a tool such as ChatGPT could save students a lot of time in preparing for assignments. He cited an example, where the tool was used to generate content for skits that students were required to perform. P11 did admit that the quality of the students' work was quite good.

Both P6 and P7 also allowed their students to use ChatGPT to generate content, which could then be critiqued as part of the assignment. Students who excelled at the assignment were the ones who were versed in the course content as they had to apply their learning in order to fulfil the requirements of the assignment. Although some participants did find opportunities to use ChatGPT in their classroom, P12 admitted that the tool was easy to use but the challenge lay in its applicability to education, when he stated, “I can't really figure out a useful way to use it that would add value. But it's ridiculously easy to use”, (P12, participant interview, October 25^th, 2024). P5 noted his concerns with this new tool as he said,

I still see it as taboo, I am struggling internally as well to accept it as being part of life and something that I can use. It creates conflict in my mind, because it goes against the traditional way of teaching and learning, where you start from the basics and you take a methodical approach. Its slowly creeping in, but it's still a fight, (P5, participant interview, May 24, 2024).

Using the social influence domain of the UTAUT framework, the data collected was organised into two major sub-themes as shown in Figure 3. Firstly, responses were allocated to an “internal” category, relating to the peers and leaders who were based within the working environment and thus “localised”. The external category was used to allocate social factors which are outside of their immediate environment, such as online forums and internationally published articles and technology trends. There were two emergent themes, which were related to social influence, leadership advocacy and student use, which will be elaborated on later in this section.

P5 disclosed that he would like to explore ChatGPT, based on what people have told him about it. However, he also re-iterated that some of the comments made by peers was that ChatGPT could be a bad thing, given that learning is supposed to be done in a particular way. This traditional approach recommends that students learn programming as a first principle, yet ChatGPT is re-arranging this where students are bypassing traditional means and getting the answers to assignments in an instant.

An interesting point was made by P8, when she stated that during their departmental meetings, the topic of ChatGPT was discussed. However, the context of these discussions was centred around ChatGPT being an “issue to be dealt with” rather than an opportunity for advancement. Other participants related their participation in workshops and training courses where they were encouraged to use it for teaching. P6 was not readily influenced by colleagues and co-workers who use the latest technology tools for teaching and learning. She emphasised that the tool should bring value to the classroom, and this is the criteria she uses for integrating new tools. The majority of persons agreed that the social factors such as colleagues' opinions would not sway them to use ChatGPT. One respondent stated that there are even lecturers who “could not be bothered” by the implication of such a tool for teaching and learning and therefore not want to change to accommodate these tools.

When asked to relate their experience of the supporting systems available to them as lecturers and instructors at the University level, there were mixed responses. P5 insisted that there is informal support; however, his administration has a traditional mindset which does not really encourage its use. Based on the responses also, the more junior participants noted that some departmental leaders were a bit apprehensive about using it, thus instilling a bit of caution in them if they were to proceed without formal approval. P5 declared that there was “no one reaching out to us” to implement AI tools such as ChatGPT. Another respondent was inclined to check out the tool but has decided to “officially” stay away from it, because of the university/leadership opinions.

An emergent theme from this study was revealed as participants not only mentioned the support of peers and leaders but also lamented the absence of leadership support and advocacy to use the tool. P5 noted that the lack of leadership to actively promote the tool was disappointing and made him cautious about using the tool on his own. P8 describes getting “little nudges from leadership” on using ChatGPT even in the absence of policy but still was unsure. P8 stated, “On my own I probably would not have really used it…I was a little worried that I didn't want my brain to get lazy…I like to be creative and think for myself too”. This theme is supported in the literature as Abdul Razzak (2015) stated that teachers can also be helped during the transition where leaders and specialists can provide moral support and reassurance. Whilst the UTAUT framework posits the effectiveness of advocacy by senior colleagues, it does not specify leadership, which emerged in this study, hence it has become a point worth mentioning.

The external sub-theme of social influence refers to those factors which are outside of participants immediate environment. It excludes colleagues and departmental teams. P11 discussed the influence of external sources that he has used. He has accessed online forums such as Reddit and articles in the Harvard Business Review to source ideas on ways to apply these AI type tools for teaching and learning. P6 also stated that she had done research on AI tools in an effort to understand it more and try to apply it in her classroom. She did get some ideas as mentioned earlier on ways to incorporate it, thus external sources such as the forums and external literature was helpful to those who were interested and wanted to learn more. Other than the sources quoted by P11 and P6, no other external sources were quoted as having a social influence on participants.

In spite of the positives that were espoused upon in terms of effectiveness, the sample did express some hesitation as it relates to using ChatGPT. P4 noted that students were “motivated by marks” and he was very cognisant of the fact that students were using it “unethically” by “using everything wholesale” meaning, they were submitting the generated output as their own. P5 concurred as he stated that “Students are using it deliberately, wildly…it is making it very, very hard to assess the quality”, (P5, participant interview, May 24, 2024).

An interesting point arose from this topic of student use and “misuse” when multiple participants surmised that the quality of education was being compromised. P6 even admitted that that is what influenced her to try to incorporate it into her teaching. She wanted to reduce the instances of misuse by students and reduce the risk of learning objectives not being met. P6 was adamant that she needed to adapt her teaching strategy to ensure there is high quality of work among students. It was essential to her to make sure that assignments forced students to “think for themselves and not depend on ChatGPT to produce responses for their work”, (P6, participant interview, May 27, 2024). P12 was very reflective when he stated that “the short-term imperative of completing assignments by the deadline crushes, the long-term imperative. Students just want to survive”. P7 was open to ideas to guard against tools which diminish the creative spirit, all the while acknowledging that “we can't pretend the students won't use it”, (P7, participant interview, May 27, 2024).

P3 believed that students were using it for “quick and easy answers” and not necessarily to learn. P3 felt that it compromised the quality of teaching and learning as the students are misusing it and they are not sure whether the answers generated are right or wrong. P5 noted that use of ChatGPT could also introduce challenges to student learning outcome, if it is misused for the assignments which are non-proctored, since they would not be adequately prepared to succeed in the proctored examination.

P7 and P12 started to give briefings to bring awareness to students on the use of the tool and the implications of using content that was not theirs when submitting assignments. P12 stated that not only should persons be aware of plagiarism, but also they should seize the opportunity to “build their intellectual muscle” in order to excel. He understood all too well the changing landscape of education as he commented “telling a student not to use ChatGPT is like telling them to go outside and not breathe”, (P12, participant interview, October 25^th, 2024).

In light of this, another point that was raised was the applicability of the tool based on the seniority of the students. P1, P4, P9, P11 and P12 all agreed that the use of ChatGPT in the classroom should not be allowed for lower-level students such as year one undergraduates, as these students need to have a foundation in the program of study prior to using ChatGPT for assignments. P12 stated children who enter an infant's classroom, are not given a pocket calculator on their first day, or even first year. They are taught the basics including foundational concepts. He likened this to “weightlifting for the brain, in order for there to be deep learning”. Both P2 and P6 were adamant when they said that the tool should not be abolished and interestingly stated that it should not be “shunned” in the classroom either.

This study was appropriately timed, as not all educators had fully embraced the new technology. They were just learning about it by attending workshops, piloting new ideas and generally trying to understand how to use the tool. Participants were aware of its current use by students; however, some were unclear as to how it could be applied in the classroom by educators. This study revealed that there is a disparity amongst educators, as there were divergent views. These views were noticeably different according to the participant's age, experience and voluntariness aligning to the moderating factors of UTAUT (gender, age, experience, voluntariness). What was also evident was the gaps which emerged between the findings and the theoretical framework, upon which the study was based as shown in Figure 4.

As the interviews progressed, it became clear that some participants who were more experienced with technology, or more “tech savvy”, had thought about the possibilities of integrating the tool in the classroom rather than shy away from it. These “tech savvy” individuals were those who liked technology and had an affinity for using it. The user-friendliness of a tool aids in its adoption (Menon & Shilpa, 2023). The data revealed that although the tool was simple and easy to use for its intended purpose as a chatbot for everyday activities, incorporating it into pedagogy was a challenge, as highlighted by participants.

Whilst ChatGPT has been adopted by users across the world in varying industries, it was not developed for use as by academia. Di Mitri et al. (2021) declared that the development of chatbots is not being driven by the needs of the learners or the pedagogical requirements in the classroom but by the technology. In other words, the advancement in the tool is not linked to the requirements of the teaching and learning process. There is a plethora of tools currently available, which were designed for use in the classroom such as Kahoot!, Canva, Flip, Padlet and ClassDojo. These are used by educators in the management of content, gamification and enhanced communication options especially for teaching and learning. Given these other options available, this study has shown that there is a lack of enthusiasm by some of the educators to use a tool such as a chatbot, for teaching.

One could argue that the technology adoption could be negatively affected due to effort expectancy, not by the ease and accessibility of the tool itself, but the effort that is required to integrate it into the teaching and learning process. On the contrary, one could argue that the diminished creativity highlighted above as a concern, is a moot point, as integration of ChatGPT does require a certain level of creativity from its users for it to be effective.

Another related component to ease of use of the tool, was the fact that educators were actually focusing on reshaping their teaching and learning strategy to combat the threat of misuse of the tool by students, rather than proactively determining how to incorporate it as a teaching tool. This discovery raises the notion that the unanimous resolve by educators, to curb the unethical use of ChatGPT, could be viewed as an indirect catalyst, incentivising educators, to actually learn how to use it, for teaching and learning. That is, rather than recoil from the reality that AI tools were being misused by students and seek to prohibit its usage, some felt compelled to indoctrinate it into their academic landscape of teaching and learning.

In spite of some of the views expressed about the uncertainty for teaching and learning, other participants revealed that they could use ChatGPT to generate “a bank of questions” to reduce repetitive content for formative assessments, assist with “mastery of content” and even suggest discussion topics to spark critical thinking. P6 also disclosed her intended strategy to deliberately ask students to generate the content to be critiqued as part of their assessment. These two examples provide evidence that the perception of the tool as being unsuitable for use in education was actually limited by the individual's knowledge, as some participants were already thinking about and piloting innovative applications of ChatGPT for teaching and learning.

There was agreement by all participants that they were not swayed simply by the social pressures to accept technology because everyone was using it, but rather the value the tool would bring to the classroom. The UTAUT framework suggests colleagues' opinions and institutional support influences users to adopt new technology. However, as mentioned in the previous section, there was a lack of visibility of what colleagues were doing and discussion amongst peers in meetings focused on ChatGPT as an “issue”. This shows that social influence in this local context was low and therefore not a key enabler to technology adoption amongst the sample population at NexTech University.

As the data were analysed, it became apparent that the lack of social influence could be attributed to lack of advocacy by senior members of staff and leaders to integrate this tool in the classroom. This is because, although it was lauded as a great tool by the senior participants, which could have value, junior participants noted the absence of leadership guidance, and only “nudges” by some senior staff in other instances. Given the lack of leadership advocacy there was a weaker social influence to use this AI tool. This aligns to Davis et al. (2003) that there is greater social impact when senior colleagues are more inclined to use the new technology, but this study extends the model to include leadership advocacy.

Although, there were some persons who indicated that they are not readily swayed by social influence on the adoption of new technology, they were intrinsically motivated to explore the options available in this case. These individuals were already researching new ways to use it on their own. Social influence would apply in instances where persons were hesitant and needed a clear leadership position before they decided whether to use it or not.

What was interesting was that the educators in the technology discipline were moving ahead with the integration of the tool in the form of “piloting” the various approaches in the absence of policy. There is a possibility that greater visibility of these “pilots” within the university has the potential to enlighten others as to the use of the tool. This would help promote its adoption via social influence, as participants said they were curious to learn about the experience of their peers in the local context. Some participants were also researching new and innovative ways to use the tool via online forums, highlighting a desire understand what was happening and change their strategy to fit.

By exploring the willingness of participants to adopt ChatGPT for teaching and learning, the research offers new insights that expand the existing theory, while challenging some of the prevailing assumptions. The results of this study highlighted that effort expectancy was not only pegged on the ease of learning the tool and using it, but also, on the practicality of the tool in the academic environment and whether it was fit for purpose. Educators were very mindful of their role and the need to ensure that learning objectives were met, therefore, they were weary of the value that an AI chatbot could bring in academia. Abdul Razzak (2015) proclaimed that learning how to use the tool, does not necessarily mean it can be integrated effectively. This lack of awareness negatively affected the perceived ease of use and ease of learning.

The sub-themes related to social influence were divided amongst the internal and external influences. Internal influences referred to peers and leaders, while external influences denoted items such as online forums, journal articles and institutional knowledge sharing websites. In addition to the internal and external sub-themes denoted by the UTAUT framework, the findings reported in this study, also pointed to the operating environment as a driving force in technology adoption. In this case, the educators are now being required to teach or deliver pedagogy in an environment where the students were already consuming content from ChatGPT. The driving force for some participants' willingness, was the students, who were their customers. The students' use/misuse was creating a demand for the educators to want to take action to resolve this situation. This customer demand within the operating environment was not clearly outlined in the UTAUT framework as an enabler of technology use. The graphic shown in Figure 5, shows the UTAUT framework together with the elements which were revealed in the study, warranting further theoretical exploration.

This case study has uncovered rich data from educators who are directly impacted and responsible for educating tertiary level students. This type of study focused on the specific context within the UTAUT model, capturing nuanced perspectives and experiences that might have been missed in broader studies. Such integration of the educational theories with the ethical and technological perspectives from educators offers a comprehensive view of this AI tool in academia. These findings can also serve as a foundational starting point for more extensive studies which can utilise quantitative analysis or comparison across institutions.

The use of purposeful sampling in this study could be considered a weakness, since the perspectives of these selected participants could have potentially skewed the results, as the opinions of the educators would be influenced by personal biases, prior experiences or varying levels of familiarity with ChatGPT. As a result, there is great potential to repeat this study to include more participants and potentially compare results across various university campuses, as part of a multiple case study design.

There was generally a consensus that the ChatGPT tool was easy to use, in its most basic form. However, there was an underlying challenge that emerged, on the use of ChatGPT in academia, as educators were somewhat perplexed by its suitability for teaching and learning. This was not deemed to be an insurmountable obstacle, as there were informal pilots underway to incorporate its application in pedagogy, by persons who were well advanced in their journey of using AI for teaching. Such experiences in the local context would bode well for others, as they can bringing awareness, by sharing the lessons learned with the rest of the faculty through their trials and experiences. Even though social influence could play a key role in adoption of this new technology, this study uncovered that the lack of formal leadership approval for the use of the tool had affected persons' willingness to engage with the tool for teaching and learning. These findings provide a potential mechanism for leaders to institutionalise AI technology in the local curriculum through policies and procedures requiring a review of the current guidelines and collaboration with educators.

The findings of this study lay a foundation for further research to be conducted with a greater sample size, which engages with more educators to understand their viewpoints. A related research topic could also be to explore the impact of various factors such as age, experience, discipline or gender in the adoption of new AI technology. Finally, it would be useful to advance the body of knowledge, by reviewing students' perceptions of AI technology tools for teaching and learning. Data collected from students would also inform policy makers and even educators to understand the viewpoints of the “customers” they serve.

A technology roadmap can also be developed to minimise risks and challenges by anticipating potential obstacles thereby reducing delays and disruptions but still ensuring that the findings of this study can be put into practice. It can also support change management, by including strategies for training, communication and gradual adoption and set clear milestones for measuring progress and success. A well-planned roadmap can help to promote long-term success by focusing not only on the initial rollout, but the ongoing support to ensure sustainability and future practice. Figures 6 and 7 show a two-phase roadmap which can be utilised to improve the willingness to adopt technology, starting with a pilot prior to the wider implementation.

This phase of the roadmap focuses on piloting the AI tool using small-scale, preliminary trials designed to evaluate the feasibility, effectiveness and uncover challenges of the AI technology. Such a program allows the institution to test the new technology in a controlled environment prior to a wider launch, allowing for feedback and data to be collected to assess the performance, identify issues and make improvements thereby minimising risks ahead of the bigger implementation. By setting clear goals for pilot programs, decision makers would have defined success metrics and methods to measure the impact.

During Phase 2, the objective would be to gradually integrate AI tools into the broader teaching practices by expanding the strategies employed during the pilot phase. Organisations who understand their user base and their perspectives of AI technology could also develop better-suited campaigns as it relates to change management. Educators in other geographies have highlighted considerations such as acceptable use policies, the ability of the tool to stimulate critical thinking, creativity and innovation, as well as integration into lesson plans would affect their decision to incorporate ChatGPT in today's classroom (Qadir, 2022). It is important that the institution provide continuous support by maintaining peer support systems and regular check-ins to address challenges.