Faculty and Institutional Readiness for Generative AI in Bangladeshi Higher Education
Abstract
Generative artificial intelligence (GenAI) is rapidly entering higher education, yet evidence from low- and middle-income countries remains limited, particularly on faculty and institutional readiness for GenAI integration. This study examined faculty and institutional readiness for generative AI in Bangladeshi higher education using a qualitative-led mixed-methods design. The primary dataset comprised 17 semi-structured interviews with faculty members from public and private universities across computer science, engineering, business, law, statistics, and social sciences, supplemented by a descriptive survey of 39 faculty members.
The findings show that faculty and institutional readiness for GenAI is uneven, socially mediated, and only weakly institutionalized. Faculty experimentation was already underway, especially among younger academics and those in STEM- and business-related fields, but confidence in critically evaluating outputs and redesigning teaching around GenAI remained limited. Demand for training was high, yet structured professional development was scarce. Governance was experienced as fragmented, with institutions relying on plagiarism detection, improvised rules, and instructor discretion rather than clear policy frameworks. Participants also described infrastructural constraints, including limited licensed access, weak learning management system integration, and reliance on informal subscription pathways.
Faculty perceptions were markedly ambivalent: GenAI was viewed as both a productivity aid and a source of concern related to academic integrity, cognitive dependency, employability, and moral responsibility. The study suggests that the central challenge lies in the uneven interim conditions under which experimentation is expanding while guidance, training, and access remain underdeveloped. These findings provide context-specific evidence for Bangladesh and comparable LMIC settings.
Keywords: generative artificial intelligence; higher education; faculty readiness; institutional readiness; governance; professional development; infrastructure; Bangladesh; low- and middle-income countries
1 Introduction
Generative artificial intelligence (GenAI) tools such as ChatGPT are reshaping higher education worldwide, influencing how students learn, how faculty teach, and how institutions govern academic work. While their efficiency and creative potential are widely praised, concerns persist about academic integrity, cognitive skill decline, and the evolving role of educators (Kohnke & Ulla, 2024; Nikolic et al., 2024). Recent analyses emphasize that readiness requires not only adoption but also the ability to critically evaluate outputs, redesign pedagogy, and embed ethical safeguards (Collie & Martin, 2024).
In high-income contexts, faculty and students often view GenAI as a mentor or productivity tool, yet simultaneously highlight risks such as plagiarism, hallucinations, and over-reliance (Chan & Hu, 2023; Wang et al., 2024). Reviews document emerging governance frameworks, integrity guidelines, and professional development programs in North America and Europe (An et al., 2025; Jin et al., 2024; Nikolic et al., 2024). More recent audits emphasize disclosure norms and fairness but also point to persistent gaps in comprehensive frameworks (An et al., 2025).
By contrast, evidence from low- and middle-income countries (LMICs) remains scarce. This creates a risk of overlooking how resource constraints, inequitable access, and weak institutional policies shape adoption (Mannuru et al., 2025; Valdivieso & González, 2025). Bangladesh is a critical case. With more than 150 universities serving millions of students, the sector faces fragile ICT infrastructure, uneven training opportunities, and fragmented governance (Yusuf et al., 2024). Adoption is nonetheless accelerating: students are widely reported to use GenAI for assignments, while faculty responses range from enthusiastic integration to strong resistance. Regional studies also highlight inequities such as reliance on free versus premium tools (Valdivieso & González, 2025) and contested perceptions of plagiarism and legitimacy (Alsharefeen & Al Sayari, 2025).
Frameworks such as the Diffusion of Innovation model and the AI Maturity Matrix highlight the roles of capabilities, training, governance, and infrastructure in shaping patterns of technology adoption and organizational readiness (Gill & Germann, 2022; Krause et al., 2025). Yet few empirical studies examine how these dimensions interact in LMICs, particularly in South Asia where higher education is expanding rapidly. Evidence on faculty readiness, training demand, and governance gaps is fragmented, leaving global debates dominated by high-income perspectives.
This study addresses these gaps by investigating faculty and institutional readiness for GenAI in Bangladeshi higher education. Using a convergent parallel mixed-methods observational design, it integrates survey data from 39 faculty members with 17 in-depth interviews. The analysis examines faculty capabilities to use and critically evaluate GenAI; training and support needs; institutional policies and governance; infrastructure readiness; faculty perceptions of risks and benefits; and variation across institutional types.
By situating Bangladesh within the global discourse on GenAI in higher education, this study contributes in three ways. First, it documents how readiness is fragmented across disciplines and generations. Second, it identifies structural inequities in infrastructure, including unofficial access pathways, rarely captured in existing models. Third, it foregrounds culturally specific framings of risk, including concerns about student “laziness” and cognitive decline, which extend debates beyond plagiarism and integrity. These contributions enrich readiness frameworks and provide context-sensitive insights for institutions and policymakers in LMICs.
2 Literature Review
The rapid uptake of generative artificial intelligence (GenAI) in higher education has generated a growing body of research on adoption patterns, governance responses, and educational impacts. While studies from high-income contexts document both opportunities and risks, evidence remains concentrated geographically. Research from low- and middle-income countries (LMICs) is comparatively limited, particularly with respect to institutional readiness, governance capacity, and equity implications. Existing literature can be synthesized across five interrelated themes.
2.1 Faculty capabilities and adoption patterns
Empirical studies consistently show uneven adoption of GenAI across disciplines and career stages. Younger academics and those in STEM-oriented fields tend to experiment earlier and more intensively than colleagues in law, humanities, and some social sciences, where ethical concerns and professional norms are more salient (Kohnke & Ulla, 2024; Maxwell et al., 2025). Beyond diffusion dynamics, motivation and institutional context shape uptake: autonomy-supportive leadership and alignment with professional development goals are associated with more active engagement (Collie & Martin, 2024). At the same time, assessment regimes and student practices increasingly exert pressure on faculty adoption, as widespread student use of GenAI reshapes instructional and evaluative expectations (Chan & Hu, 2023).
However, recent work cautions against overgeneralizing demographic predictors. Once institutional context and governance conditions are accounted for, age and discipline effects may weaken or vary (Nevárez Montes & Elizondo-Garcia, 2025). Importantly, most studies focus on frequency of use rather than faculty capability to critically evaluate GenAI outputs, an omission that is especially pronounced in LMIC research. As a result, the literature provides an incomplete picture of readiness beyond surface-level adoption.
2.2 Training and support needs
Across contexts, professional development is identified as a prerequisite for responsible GenAI integration. Faculty consistently express demand for structured training but emphasize that effective programs must integrate pedagogical design, ethical reasoning, and critical evaluation alongside technical tool literacy (Kohnke & Ulla, 2024; Krause et al., 2025; Nikolic et al., 2024; Shata & Hartley, 2025). Generic or tool-centric training is often viewed as insufficient or misaligned with disciplinary values.
In LMIC settings, including Bangladesh, a recurring gap emerges between interest and provision. Studies report high levels of curiosity and perceived relevance alongside limited access to formal training opportunities (Mannuru et al., 2025; Yusuf et al., 2024). This disconnect suggests that readiness is constrained less by motivation than by the absence of locally feasible, institutionally supported models of professional development.
2.3 Institutional policies and governance
Institutional governance responses to GenAI are evolving but remain fragmented. Policy scans and reviews from high-income contexts describe an “open but cautious” stance, where guidance documents, workshops, and syllabus statements proliferate faster than binding regulations (Chan & Hu, 2023; Jin et al., 2024; Wang et al., 2024). Comparative analyses highlight path-dependent governance models shaped by legal traditions, academic culture, and state involvement (Li et al., 2025).
For LMICs, emerging proposals emphasize layered governance arrangements that combine institutional guidance, disclosure norms, and academic integrity safeguards while accounting for resource constraints (Alli et al., 2025). Despite growing attention, reviews continue to note the absence of comprehensive, enforceable frameworks and a lack of practical guidance tailored to constrained institutional environments (An et al., 2025; Nikolic et al., 2024). Governance in such contexts is often reactive, relying on plagiarism detection tools and individual instructor discretion rather than coordinated policy.
2.4 Infrastructure and resource gaps
Infrastructure constitutes a foundational dimension of GenAI readiness. Studies from LMICs consistently document deficits in connectivity, device availability, power reliability, and access to licensed software, constraints that limit adoption regardless of faculty interest or skill (Al-kumaim et al., 2025; Mannuru et al., 2025). Research from Africa and South Asia further details how these deficits translate into practice, including reliance on smartphones over laptops, bandwidth-sensitive usage patterns, and pronounced rural–urban disparities (Kumar, 2025; Modiba et al., 2025; Mutelo, 2025; Valdivieso & González, 2025).
Such conditions shape not only whether GenAI is used, but how it is used. Without institutional investment and trustworthy access pathways, adoption may reinforce existing inequalities within and between institutions, challenging assumptions embedded in readiness models developed in high-resource settings.
2.5 Perceptions of GenAI: opportunities and risks
Faculty and student perceptions of GenAI are consistently ambivalent. Across contexts, benefits are associated with drafting support, feedback, and idea generation, while concerns center on plagiarism, hallucinations, and erosion of higher-order cognitive skills (Chan & Hu, 2023; Nikolic et al., 2024; Shata & Hartley, 2025). Importantly, these perceptions are mediated by cultural, moral, and disciplinary lenses.
Studies from the Gulf and other regions illustrate differentiated understandings of AI-assisted misconduct and a preference for educative rather than punitive responses (Alsharefeen & Al Sayari, 2025). African and multicultural research foregrounds integrity, privacy, equity, and professional identity alongside pedagogical adaptability (Pramjeeth & Ramgovind, 2024; Yusuf et al., 2024). These findings underscore that risk perceptions are not solely technical but embedded in local value systems, reinforcing the need for context-sensitive governance and pedagogy.
2.6 Gaps and this study’s contribution
Despite rapid growth in the literature, three gaps persist. First, empirical evidence from LMICs remains limited, constraining the generalizability of readiness models derived largely from high-income contexts. Second, existing studies rarely examine how disciplinary resistance, training uptake, and governance conditions interact under resource constraints. Third, structural inequities in access—such as smartphone-dominated use, limited LMS integration, and informal subscription pathways—are seldom integrated into readiness frameworks.
By jointly examining faculty capabilities, training demand, governance awareness, infrastructure conditions, and culturally inflected risk narratives within a single empirical design, this study extends GenAI readiness research with LMIC-specific evidence from Bangladesh. In doing so, it highlights practical priorities for ethics-integrated professional development, disclosure-oriented governance, and equitable infrastructure provisioning in resource-constrained higher education systems.
3 Methods
3.1 Research Design
This study employed an exploratory qualitative-led design to examine how faculty members in Bangladesh perceive readiness for generative artificial intelligence (GenAI) integration in higher education. The primary emphasis was qualitative inquiry, as readiness was understood not only as a technical or organizational condition but also as a socially interpreted phenomenon shaped by disciplinary norms, institutional cultures, and perceived risks.
To provide supplementary contextual evidence, a small faculty survey was conducted alongside the interviews. The survey was intended to capture descriptive patterns in self-reported GenAI use, access to training, awareness of governance arrangements, and perceived infrastructural conditions. It was not designed for statistical generalization or inferential testing. Quantitative and qualitative materials were therefore used asymmetrically: interviews formed the principal analytic basis of the study, while survey findings served as supporting descriptive context.
3.2 Participants
Participants were recruited from public and private universities in Bangladesh using purposive and snowball sampling. Eligibility was restricted to faculty members currently holding teaching or research positions at recognized higher education institutions in Bangladesh. Recruitment was designed to capture disciplinary variation and a range of orientations toward GenAI, including likely adopters, cautious users, and more skeptical faculty. Initial recruitment points were placed across multiple departments and institutional types to improve diversity within the sample.
The primary qualitative sample comprised 17 faculty members who participated in semi-structured interviews. Participants represented computer science, engineering, business, law, statistics, social sciences, and related fields. They varied in age, career stage, and institutional setting, allowing comparison across disciplinary and generational contexts. Anonymized identifiers were used throughout the analysis.
A supplementary descriptive survey was completed by 39 faculty members. Most respondents were affiliated with public universities, with a smaller proportion from private institutions. Because the survey sample was small and nonprobabilistic, it was used for contextual orientation rather than representative estimation.
3.3 Survey Instrument
The survey instrument was developed to capture contextual indicators relevant to GenAI readiness in Bangladeshi higher education. It included items on self-reported GenAI use, perceived confidence and capability, access to training and support, awareness of institutional policies, infrastructural conditions, and perceived opportunities and risks.
The instrument contained a combination of five-point Likert-type items, multiple-response questions, and a limited number of open-text responses. Different response formats were used because some items were intended to capture attitudes, whereas others focused on practices, access conditions, or perceptions of institutional arrangements.
The survey was designed for descriptive use within an exploratory study rather than as a psychometrically validated readiness scale. To improve wording clarity and contextual fit, the instrument was reviewed and pilot-tested with three faculty members prior to administration.
3.4 Interview Protocol
The semi-structured interview guide was designed to generate in-depth accounts of how faculty interpreted GenAI adoption, institutional preparedness, and related risks. Core topics included experiences using or resisting GenAI in teaching and research; views on training and professional development; perceptions of institutional governance and policy clarity; access to relevant infrastructure and software; and broader judgments about the educational, ethical, and professional implications of GenAI.
Interviews lasted approximately 30 to 50 minutes and were conducted either in person or online, depending on participant preference and availability. Interviews were conducted in Bangla or English. With participant consent, interviews were audio-recorded, transcribed, anonymized, and translated into English where necessary for reporting. Care was taken to preserve the original meaning and tone of participant accounts.
3.5 Data Collection Procedures
Data collection was conducted in August 2025. Survey invitations were distributed through email and professional networks. Participation was voluntary and uncompensated. Interview participants were recruited through survey follow-up and professional contacts using purposive and snowball strategies.
Consent procedures were adapted to the mode of participation. Written consent was obtained for in-person interviews, whereas verbal consent was obtained and audio-recorded for online interviews. All participants were informed of the purpose of the study, the voluntary nature of participation, and the measures taken to protect confidentiality.
3.6 Data Analysis
The qualitative data formed the primary analytic core of the study. Interview transcripts were analyzed using thematic analysis following Braun and Clarke’s reflexive approach (Braun & Clarke, 2006). Initial coding was informed by the study focus areas, including capability, training, governance, infrastructure, and perceptions of risk and benefit. Codes and candidate themes were refined iteratively as additional patterns emerged during close reading of the transcripts, including unofficial access pathways, concerns about student dependency, and culturally grounded moral narratives.
Two researchers independently engaged with the transcripts and discussed evolving interpretations, coding decisions, and candidate themes through iterative comparison. Reflexive memos were maintained throughout the analytic process to document interpretive decisions and monitor assumptions.
Survey data were analyzed descriptively using frequencies and percentages. These findings were used to contextualize the interview results rather than to test hypotheses or establish generalizable prevalence estimates. Integration occurred through a qualitative-led interpretive strategy in which survey patterns were considered alongside interview accounts to identify areas of convergence, contrast, and contextual elaboration.
3.7 Methodological Limitations
This study was exploratory and based on nonprobability sampling. The findings therefore should not be interpreted as statistically representative of all universities or faculty members in Bangladesh. Participation may also have been shaped by self-selection, with greater involvement from those already engaged with or concerned about GenAI.
In addition, the survey component was descriptive and not based on a validated readiness scale. The sample of survey respondents was weighted toward public university participants, limiting robust institutional comparison. The value of the study lies not in population-level estimation, but in its ability to illuminate how readiness is being understood, experienced, and negotiated within a specific higher education context.
3.8 Ethical Considerations
Formal approval from an institutional research ethics committee was not required under applicable national regulations or prevailing institutional practice for this type of noninterventional social science research involving adult participants in Bangladesh. The study was conducted in accordance with recognized ethical principles for human-participant research.
All participants received information about the purpose of the study, the voluntary nature of participation, and the handling of confidentiality before data collection began. Interview transcripts and survey data were anonymized, stored securely, and used solely for research purposes. In reporting the findings, identifying details were minimized or generalized to reduce the risk of re-identification.
4 Results
This section presents the findings as a qualitative-led account of how faculty members interpret readiness for generative AI in Bangladeshi higher education. The semi-structured interviews provide the primary analytic basis of the section, while the survey results are used descriptively to contextualize recurring patterns in reported use, training access, governance awareness, infrastructural conditions, and perceived opportunities and risks. Findings are organized thematically across six interrelated areas: faculty capability and use, training and support, governance, infrastructure, perceptions, and institutional variation. Supporting descriptive survey indicators and illustrative interview excerpts are provided in Tables 1–3.
4.1 Participant Profile
The primary qualitative sample comprised 17 faculty members from public and private universities in Bangladesh. Participants represented a range of disciplinary backgrounds, including computer science, engineering, business, law, statistics, and social sciences, and varied in age, career stage, and institutional context. This diversity enabled comparison across disciplinary cultures and institutional settings.
To provide descriptive context, a supplementary survey was completed by 39 faculty members. Most survey respondents were employed at public universities, and the sample was predominantly male. Nearly four fifths had fewer than ten years of teaching experience. Because the survey sample was small and non-probabilistic, these figures are presented as contextual indicators rather than representative estimates of faculty populations in Bangladesh.
Table 1. Descriptive Profile of Survey Respondents
| Category | Subcategory | Count (n) | % |
| Institution Type | Public | 32 | 82.1 |
| Private | 7 | 17.9 | |
| Years of Teaching Experience | < 5 years | 15 | 38.5 |
| 5–10 years | 16 | 41.0 | |
| 10–20 years | 3 | 7.7 | |
| > 20 years | 5 | 12.8 | |
| Gender | Male | 34 | 87.2 |
| Female | 5 | 12.8 |
Legend: Table 1 presents descriptive characteristics of the survey respondents (n = 39). These figures are included to contextualize the qualitative findings and should not be interpreted as representative of the wider faculty population in Bangladesh.
4.2 Uneven Capability and Selective Adoption
A recurring pattern across the interviews was the uneven relationship between familiarity with generative AI tools and deeper pedagogical or evaluative capability. Participants often distinguished between being able to operate tools such as ChatGPT and being able to critically assess their outputs, judge their reliability, or integrate them meaningfully into teaching. The survey results broadly aligned with this pattern: a majority of respondents reported confidence in using GenAI tools, yet far fewer indicated confidence in evaluating output quality or designing AI-supported learning activities.
Interview accounts suggested that this gap was shaped by disciplinary norms, trust in the technology, and perceived professional risk. Faculty in STEM- and business-related fields, especially those earlier in their careers, more often described GenAI as a productivity tool that accelerated drafting, idea generation, and routine academic tasks. One participant expressed this efficiency-oriented perspective directly: “To be honest, what would normally take me a week, AI can do for me in five minutes” (B3, STEM, private university).
Other participants adopted a more restricted orientation. Faculty in law and some social science fields described GenAI as potentially useful for brainstorming or preliminary exploration but inappropriate for core academic writing, assessment, or disciplinary reasoning. In these accounts, acceptable use was bounded by concerns about authorial responsibility, academic integrity, and professional credibility. A further group described selective or declining use because of doubts about the accuracy and consistency of outputs. These participants referred to hallucinations, misleading responses, or weak contextual relevance, which reduced their willingness to rely on GenAI in teaching or research.
Taken together, the findings suggest that readiness cannot be inferred from use alone. Faculty may experiment with GenAI while still lacking confidence in critical evaluation or pedagogical integration, and similar levels of exposure can lead to markedly different practices across disciplines.
4.3 Strong Demand for Training, but Not Consensus on Training
Training emerged as one of the most visible dimensions of readiness, but also one of the most contested. Many interview participants described the need for structured professional development that would address not only tool use, but also pedagogy, assessment redesign, ethics, and responsible classroom integration. This pattern was reflected in the survey, where a large majority of respondents indicated that structured training would improve their ability to use GenAI effectively, while only a small minority reported that their institution currently offered such training.
However, interview accounts showed that support for training was not uniform. Some participants regarded training as essential for both faculty and students, particularly if institutions were to establish shared expectations and reduce confusion about acceptable use. As one faculty member noted, “Training is needed in both domains, teaching staff and students” (A12, MIS, public university). These participants viewed capacity-building as a precondition for responsible adoption rather than as an optional enhancement.
Others were more conditional in their support. They expressed willingness to engage in training only if it were discipline-sensitive, led by credible facilitators, and oriented toward real classroom dilemmas rather than generic enthusiasm for AI tools. In these accounts, superficial workshops were seen as unlikely to help faculty navigate the substantive pedagogical and ethical questions raised by GenAI.
A smaller but important group resisted the idea of training more fundamentally. Their concern was not merely that training was unavailable, but that training itself might signal institutional endorsement of practices they considered professionally or ethically problematic. One law participant stated, “Because the effects are mostly negative, I do not think training is necessary” (A11, Law, public university). This suggests that training is not simply a matter of institutional provision. It is also interpreted symbolically, and in some disciplines it becomes entangled with broader moral and professional concerns.
In the absence of formal institutional provision, many participants described relying on self-directed experimentation, peer advice, and informal learning networks. While these pathways supported some degree of adaptation, they were uneven and rarely accompanied by explicit guidance on pedagogy, assessment, or ethics.
4.4 Governance as Improvised and Uneven
The interviews consistently portrayed governance as fragmented, improvised, and often devolved to the level of individual instructors. Participants rarely described comprehensive institutional frameworks for GenAI use. Instead, they referred to inherited academic integrity rules, informal verbal expectations, and the use of plagiarism detection or AI-detection tools as partial substitutes for explicit policy. Survey responses pointed in the same direction, with only a small proportion of respondents reporting clear institutional rules and many indicating the absence of structured policies.
Several participants described this policy vacuum directly. One faculty member stated, “There are no policies or guidelines in my institution, only plagiarism detection software” (A11, Law, public university). This comment reflected a broader pattern in which technical monitoring tools were often treated as proxies for governance, even though participants also expressed doubts about their reliability and fairness.
In the absence of formal policy, faculty described constructing their own working norms. These included expectations that students disclose GenAI use, informal thresholds for what might be considered acceptable assistance, and decisions to discourage AI use in high-stakes assignments. Yet these norms varied substantially across instructors and departments. As a result, students were likely to encounter inconsistent expectations depending on course context, discipline, and faculty disposition.
Some participants emphasized transparency and disclosure, arguing that AI-assisted work should be openly acknowledged rather than hidden. Others focused on restriction and control, including the use of detectors or examination rules. Still others relied on pre-existing plagiarism frameworks without distinguishing clearly between plagiarism and disclosed AI-assisted writing. Across these accounts, governance appeared less as a stable institutional structure than as a set of improvised responses to uncertainty.
Together, these accounts indicate that faculty concern about GenAI often coexists with limited institutional coordination. Readiness in this area is therefore constrained not only by the absence of formal policy, but also by the inconsistent translation of concern into shared, workable, and communicable expectations.
4.5 Partial and Informal Infrastructure
Interview accounts suggested that infrastructural readiness was partial rather than absent. Many participants had some form of access to GenAI through personal devices, shared computer laboratories, or free-tier services, but this access was often fragile, uneven, and weakly embedded in institutional systems. Survey responses were broadly consistent with this picture: while some respondents judged their institution’s infrastructure as supportive of GenAI use, learning management system integration was rare, and a substantial minority considered existing infrastructure inadequate.
Participants described infrastructure challenges at several levels. At the material level, they referred to crowded laboratories, limited numbers of functional computers, and inadequate access to higher-end computing resources. One participant remarked, “In our lab there are 50 students but only 40 computers” (A12, MIS, public university), while another noted the absence of sufficient GPU capacity for more advanced AI-related work (A9, STEM). These limitations constrained what faculty imagined as realistically possible in classroom practice.
Infrastructure was also discussed as an access ecology involving software, subscriptions, and platform integration. Many participants reported that GenAI use depended on personal accounts, free services, or informal premium access rather than institutionally licensed and governed pathways. In some accounts, both faculty and students were said to rely on unofficial or resale access to paid tools. This raised concerns not only about equity, but also about privacy, sustainability, and institutional control.
The interviews also pointed to a distinction between functional access and broader institutional readiness. Faculty may be able to use GenAI individually, but such use does not amount to stable institutional capacity if it depends on personal hardware, informal subscriptions, and disconnected workflows. Participants frequently expressed interest in institution-level licensing, better platform integration, and more coordinated planning, but these aspirations often exceeded current procurement and resource realities.
4.6 Ambivalent Perceptions of Opportunity and Risk
Faculty perceptions of GenAI were consistently ambivalent. Participants rarely described the technology as either wholly beneficial or wholly harmful. Instead, they positioned it as a tool with visible utility but also substantial educational, ethical, and social risks. Survey responses mirrored this ambivalence, with many respondents expressing openness to experimentation and perceived teaching benefits, while substantial proportions also emphasized academic integrity concerns and the possible erosion of human teaching roles.
One common framing was that GenAI could function as an assistant, mentor, or productivity aid when kept under human control. In this view, it could support drafting, feedback, and preparation without displacing academic judgment. As one participant explained, “I can use AI as my mentor, but I cannot follow it 100%” (A8, STEM). This perspective was especially visible among those who supported conditional or supervised adoption.
At the same time, many participants expressed concern about what GenAI might do to student habits of learning. A recurring fear was that students would become overly dependent on machine-generated output, leading to weaker writing, reduced reasoning effort, and diminished intellectual discipline. One participant stated, “Our students have become lazy, their writing capacity will decline” (A13, Law/Business). These concerns extended beyond misconduct in a narrow sense and touched on broader anxieties about educational formation.
Some participants also interpreted GenAI through moral, professional, or religious lenses. In these accounts, undisclosed reliance on AI-generated content was not simply a procedural breach, but a broader ethical failure. Others connected GenAI to labour market uncertainty and worried that automation would reduce opportunities for graduates, thereby reshaping the social role of higher education.
Taken together, the findings show that faculty interpretations of readiness were shaped by professional and culturally situated narratives. Faculty do not simply assess the usefulness of a tool; they interpret it in relation to ideas about diligence, integrity, expertise, and social responsibility.
4.7 Variation Across Discipline, Career Stage, and Institution
Participants often perceived differences between public and private universities in terms of flexibility, resources, and openness to GenAI. Survey responses likewise suggested that many respondents viewed private universities as more advanced in this area. However, the interview findings indicated that sector alone did not explain the most meaningful differences in practice.
Two other dimensions appeared more salient: discipline and career stage. Across both public and private institutions, STEM-, MIS-, and business-related departments were more often associated with experimentation, practical use, and openness to integration. By contrast, law and some social science fields more frequently expressed caution, principled resistance, or stronger concern about ethical and professional implications.
Generational patterns were also evident in participant accounts. Early-career and mid-career faculty tended to report more frequent experimentation and greater ease with digital tools, while more senior faculty were often more cautious or selective in their use. These differences were not absolute, but they recurred across interviews often enough to suggest that readiness is experienced unevenly within institutions rather than simply between sectors.
This layered pattern complicates simple narratives of private-sector advancement or public-sector lag. Institutional type may shape background conditions, but day-to-day readiness appears to be negotiated more directly at the intersection of disciplinary norms, career stage, and perceived risk.
Table 2. Descriptive Survey Indicators Used to Contextualize Interview Themes
| Theme | Survey Item | Count (n) | % |
| Capability and Use | Confident in using GenAI tools | 24 | 61.5 |
| Use GenAI in teaching/research | 17 | 43.6 | |
| Confident in evaluating accuracy/bias of outputs | 11 | 28.2 | |
| Confident in designing teaching activities with AI | 11 | 28.2 | |
| Training and Support | Institution offers GenAI-related training | 4 | 10.3 |
| Structured training would help | 34 | 87.2 | |
| Librarian/IT support would help | 21 | 53.8 | |
| Governance | Clear policies exist | 5 | 12.8 |
| Aware of ethical guidelines | 20 | 51.3 | |
| Expectations/consequences are communicated | 7 | 17.9 | |
| No structured policies | 16 | 41.0 | |
| Infrastructure | Infrastructure supports GenAI use | 21 | 53.8 |
| LMS integrates GenAI | 3 | 7.7 | |
| Institution invests in upgrades | 13 | 33.3 | |
| Infrastructure is inadequate | 14 | 35.9 | |
| Perceptions | GenAI can improve teaching/learning | 25 | 64.1 |
| GenAI poses risks to academic integrity | 16 | 41.0 | |
| Open to experimenting with GenAI | 24 | 61.5 | |
| Concerned about AI replacing teaching roles | 11 | 28.2 | |
| Institutional Variation | Private universities are more advanced | 28 | 71.8 |
| My university is more prepared | 6 | 15.4 | |
| My department has taken steps | 6 | 15.4 | |
| I do not know | 4 | 10.3 |
Legend: Table 2 summarizes descriptive survey indicators used to contextualize the qualitative findings. Percentages should be interpreted cautiously because the survey was exploratory, non-probabilistic, and included a small private-university subsample.
Table 3. Illustrative Interview Excerpts by Theme
| Theme | Illustrative Quote (English translation) | Faculty ID |
| Capability and Use | “To be honest, what would normally take me a week, AI can do for me in five minutes.” | B3 (STEM, private university) |
| Capability and Use | “Our generation, and especially the next, is using this much more heavily.” | A13 (STEM, public university) |
| Capability and Use | “In research, I might take an idea from AI, but the writing must always be in my own words.” | A11 (Law, public university) |
| Training and Support | “Training is needed in both domains, teaching staff and students.” | A12 (MIS, public university) |
| Training and Support | “If sessions were organized with good resource persons, faculty would respond positively.” | A10 (Statistics, public university) |
| Training and Support | “Because the effects are mostly negative, I do not think training is necessary.” | A11 (Law, public university) |
| Governance | “There are no policies or guidelines in my institution, only plagiarism detection software.” | A11 (Law, public university) |
| Governance | “AI-generated content should be disclosed. Otherwise, it becomes an ethical issue.” | A13 (Law/Business, public university) |
| Governance | “Not more than 15% [AI content] should be allowed.” | B1 (STEM, private university) |
| Infrastructure | “In our lab there are 50 students but only 40 computers.” | A12 (MIS, public university) |
| Infrastructure | “There is a GPU gap, we do not have the resources for deep learning.” | A9 (STEM, public university) |
| Infrastructure | “If we had integrated AI software across the faculty, we could save much more time.” | B1 (STEM, private university) |
| Perceptions | “I can use AI as my mentor, but I cannot follow it 100%.” | A8 (STEM, public university) |
| Perceptions | “Our students have become lazy, their writing capacity will decline.” | A13 (Law/Business, public university) |
| Institutional Variation | “In my experience, private universities are more advanced in AI adoption.” | A12 (MIS, public university) |
| Institutional Variation | “Companies are no longer recruiting fresh graduates. This is alarming.” | A9 (STEM, public university) |
Legend: Table 3 presents illustrative interview excerpts aligned with the major themes in the analysis. All excerpts were translated into English where necessary and attributed to anonymized participant identifiers. Disciplinary references are included to preserve contextual variation while maintaining anonymity.
5 Discussion
5.1 Framing readiness: experimentation, pacing, and the interim period
This study examined faculty and institutional readiness for generative AI integration in Bangladeshi higher education using a qualitative-dominant design supported by descriptive survey data. Across the findings, a consistent pattern emerged: faculty experimentation is already underway, but it is unfolding within an institutional environment where governance, training, and access remain uneven. This asymmetry should not be interpreted automatically as institutional failure. As with other emerging technologies, slower institutional response may reflect caution, deliberation, and the need to avoid premature formalization of practices that are still poorly understood (An et al., 2025; UNESCO, 2023; Wang et al., 2024).
In the present study, however, the central issue appears to be the character of the interim period rather than institutional pacing alone. Faculty are already using GenAI in selective and sometimes innovative ways, yet shared guidance, structured professional support, and stable access arrangements remain underdeveloped. As a result, readiness is not absent, but fragmented. It is being negotiated locally through individual practice, disciplinary norms, and informal workarounds rather than through coherent institutional arrangements. The discussion below interprets this pattern across governance, infrastructure, capabilities and training, perceptions, and institutional variation.
5.2 Institutional policies and governance
The findings suggest that governance readiness for GenAI in Bangladeshi higher education remains weakly formalized. More than two-fifths of survey respondents reported the absence of any structured GenAI policy, and only a small minority perceived clear institutional expectations. Although just over half indicated awareness of some ethical guidance, interview accounts showed that these were usually generic academic integrity rules rather than GenAI-specific frameworks. Under these conditions, responsibility for determining acceptable practice is often devolved to departments and individual faculty.
Importantly, this does not mean that governance is absent in practice. Rather, governance appears to be improvised. Faculty described recurring responses such as extending plagiarism rules and detection systems to GenAI without formal adaptation, constructing informal disclosure norms or usage thresholds, and relying on procedural controls such as phone restrictions during examinations. These responses are motivated by integrity concerns, but they operate unevenly and produce inconsistent expectations across courses and institutions. This pattern is broadly consistent with international evidence showing that many universities remain in an early stage of GenAI governance development, often relying on partial guidance, fragmented institutional responses, or evolving policy documents rather than comprehensive and stable (An et al., 2025; Jin et al., 2024; Wang et al., 2024).
At the same time, the Bangladeshi case appears to differ in degree and institutional depth. In many higher-income settings, an “open but cautious” stance is accompanied by formal resources, guidance documents, or institutional statements that provide at least some shared reference points (UNESCO, 2023; Wang et al., 2024). In this study, by contrast, improvisation often substitutes for policy rather than supplementing it. Faculty also expressed limited trust in AI-detection tools, raising concerns about fairness, reliability, and the risks of false accusation. This concern echoes recent work showing that academic integrity practices in the AI era remain unsettled and that faculty often face uncertainty when translating policy principles into workable classroom procedures (Alsharefeen & Al Sayari, 2025; Yusuf et al., 2024).
Taken together, these findings suggest that governance readiness in resource-constrained contexts should not be evaluated solely in terms of whether institutions respond quickly. A slower institutional response may be reasonable. The more immediate challenge is whether institutions provide workable interim guidance while faculty experimentation is already expanding. In that setting, clear disclosure norms, feasible assessment guidance, and locally realistic enforcement processes appear more important than attempts to govern GenAI primarily through detection. The findings therefore support emerging calls for governance approaches that are educative, context-sensitive, and institutionally realistic rather than narrowly punitive (Alli et al., 2025; Li et al., 2025; UNESCO, 2023).
5.3 Infrastructure readiness
Infrastructure emerged in this study as both an enabler and a limiting condition for GenAI integration. While just over half of survey respondents perceived their institution’s infrastructure as supportive, learning management system integration was rare and more than one-third judged existing infrastructure inadequate. Interview data clarified that such constraints shape not only whether GenAI is used, but how it is used. Overcrowded laboratories, outdated machines, and limited access to advanced computing resources restricted the range of possible academic applications, often confining GenAI use to basic prompting rather than deeper curricular or research integration.
Participants also described infrastructure as an access ecosystem rather than a simple matter of hardware. Faculty referred to combinations of personal devices, shared institutional computers, free-tier services, plagiarism detection systems, and informal premium-tool access. This suggests that readiness depends not only on technological availability, but also on the legitimacy, stability, and governability of access pathways. When access depends on shared logins, unofficial subscriptions, or informal resale arrangements, institutional readiness remains fragile even if faculty can technically reach the tools.
This finding extends current work on GenAI in developing and resource-constrained settings, which often emphasizes connectivity gaps, digital inequality, and uneven technological preparedness (Kumar, 2025; Mannuru et al., 2025; Valdivieso & González, 2025). Evidence from African and Global South contexts similarly suggests that infrastructural barriers are not limited to bandwidth or devices, but include affordability, platform dependence, sustainability, and unequal institutional support (Modiba et al., 2025; Mutelo, 2025). The present study adds that these issues also involve what may be termed access integrity: whether access is equitable, legitimate, secure, and aligned with institutional systems.
Seen in this way, infrastructural readiness for GenAI in higher education should not be reduced to device ownership or internet availability alone. It also includes licensing legitimacy, platform integration, identity management, procurement capacity, and the ability of institutions to provide stable and trustworthy access environments. The key issue is therefore not merely whether institutions move quickly to install AI-related systems, but whether they can create credible and sustainable access conditions while adoption is already unfolding.
5.4 Faculty capabilities and training
The findings show a clear distinction between operational familiarity with GenAI and deeper pedagogical readiness. Although many respondents reported confidence in using GenAI tools and some had already integrated them into teaching or research, far fewer felt able to evaluate outputs critically or design meaningful pedagogical applications. This suggests that surface-level use may diffuse more rapidly than the evaluative and instructional capacities needed for responsible integration. That distinction is consistent with emerging literature showing that faculty adoption of GenAI is shaped not only by perceived usefulness, but also by motivation, contextual supports, confidence, and discipline-specific interpretations of relevance (Collie & Martin, 2024; Nevárez Montes & Elizondo-Garcia, 2025; Nikolic et al., 2024).
Training emerged as the most widely endorsed institutional need, yet also as one of the most contested. Survey responses suggested strong support for structured professional development, but interviews revealed that faculty do not interpret training in uniform ways. Some participants actively sought training that could connect GenAI to pedagogy, ethics, and critical evaluation. Others were willing to engage only if such training was discipline-sensitive, credible, and grounded in real academic challenges. A smaller group resisted training because they viewed it as legitimizing a technology they considered professionally risky or ethically problematic.
This is an important nuance. Existing literature often treats professional development as a natural or broadly accepted response to emerging educational technologies (Kohnke & Ulla, 2024; Shata & Hartley, 2025). The present findings suggest that in contested domains such as GenAI, resistance may be directed not only at the tools themselves, but also at the institutional act of formalizing support around them. Training is therefore not always perceived as a neutral capacity-building intervention. It can also function symbolically, signaling what an institution is prepared to normalize.
At the same time, participants described active informal learning through experimentation, peer exchange, and self-directed exploration. These pathways appear to sustain adoption in the absence of formal support, but they are uneven and do not necessarily provide adequate grounding in ethics, evaluation, or instructional design. Here again, the issue is not simply that institutions have moved slowly. Rather, the problem is that the interim period between early experimentation and more formal institutional response is weakly supported. Professional development, therefore, should not merely teach tool use. It should address disciplinary concerns, strengthen critical judgment, and build on existing peer-learning practices while avoiding one-size-fits-all models.
5.5 Perceptions: benefits, risks, and local framings
Faculty perceptions of GenAI in this study were marked by ambivalence rather than simple optimism or rejection. Many participants described GenAI as helpful for drafting, idea generation, feedback, or efficiency, while also expressing concern about academic integrity, declining student effort, weakened writing and reasoning skills, and the possible erosion of human teaching roles. This broad pattern is consistent with international literature showing that GenAI is often simultaneously viewed as beneficial and risky within higher education (Chan & Hu, 2023; Krause et al., 2025; Yusuf et al., 2024).
However, the qualitative data suggest that such ambivalence is not merely a matter of balancing pros and cons. It is also shaped by culturally situated interpretive frames. Some participants conceptualized GenAI as a subordinate assistant or “mentor” that could be used under careful human control. Others linked it to anxieties about laziness, cognitive dependency, loss of authenticity, or threats to professional and moral standards. In several accounts, concerns about GenAI extended beyond academic performance to broader issues such as employability, social change, and moral responsibility.
These interpretations resonate with recent scholarship arguing that AI adoption in education cannot be understood only through technical capability or policy readiness, but must also account for ethical, social, and normative framings (Gill & Germann, 2022; Pramjeeth & Ramgovind, 2024). The present study suggests that in the Bangladeshi context, these framings are central to readiness. GenAI is not encountered simply as a productivity tool. It is interpreted through local values related to diligence, accountability, credibility, and the social purpose of higher education.
This matters because such perceptions shape practice. Faculty who framed GenAI as a manageable support tool were more likely to explore conditional integration. Those who framed it as morally corrosive, professionally unsafe, or harmful to student development were more likely to discourage or restrict use. Readiness, therefore, is not only a function of infrastructure, skills, or policy clarity. It is also mediated by locally meaningful beliefs about what counts as good teaching, responsible scholarship, and legitimate academic effort.
5.6 Institutional variation and analytical contributions
Although survey respondents often perceived private universities as more advanced in GenAI integration, the broader pattern in this study suggests that institutional variation is more layered than a simple public-private divide. Interview data indicated meaningful variation within both sectors, with disciplinary culture and career stage appearing more influential than sector alone in shaping attitudes and practices. STEM, business, and MIS-related participants were generally more open to experimentation, whereas law and some social science participants expressed stronger reservations. Younger and mid-career faculty also appeared more likely to report active exploration than more senior colleagues.
This interpretation is broadly consistent with recent literature showing that GenAI adoption in higher education varies across context, discipline, demographic background, and institutional culture rather than following a single uniform trajectory (Collie & Martin, 2024; Maxwell et al., 2025; Nikolic et al., 2024; Shata & Hartley, 2025). The current study suggests that sector may shape background conditions such as resources, expectations, or aspirational narratives, but everyday readiness is negotiated more locally through disciplinary norms, generational experience, and perceived risk.
From an analytical standpoint, the study offers several contributions. First, it suggests that resistance may target not only the technology itself, but also the training and institutionalization processes surrounding it. Second, it highlights access integrity as an important dimension of infrastructural readiness in resource-constrained settings. Third, it shows that cultural and moral narratives are not peripheral concerns; they actively structure how faculty interpret GenAI and its place in academic life. Fourth, and more broadly, the study suggests that the key challenge may not be institutional slowness alone, but the governance and support conditions that characterize the period between early experimentation and formal institutional adaptation. These points do not produce a universal readiness model, but they do help refine current discussions of GenAI adoption by foregrounding contextual and interpretive dimensions that may be especially important in LMIC higher education environments (Mannuru et al., 2025; Valdivieso & González, 2025).
5.7 Limitations
This study has several limitations that should be considered when interpreting the findings. First, the survey and interview participants were recruited through purposive and snowball approaches, which limits representativeness and may have amplified the perspectives of faculty who were already engaged with, concerned about, or interested in GenAI. Second, the survey sample was small and skewed toward public university faculty, which constrains the strength of sectoral comparison and means that the quantitative results should be interpreted as descriptive rather than generalizable.
Third, the study relied on self-reported accounts of readiness, use, and institutional conditions. Such accounts are valuable for understanding perception and meaning, but they may also be influenced by recall limitations, self-presentation, or unequal familiarity with institutional processes. Fourth, the sample was predominantly male, which means the study may not adequately reflect gendered dimensions of GenAI perception and adoption in Bangladeshi higher education.
Fifth, although broader institutional culture may shape how GenAI practices move from individual experimentation to formal response, this study did not directly examine communication dynamics between junior faculty, senior faculty, and institutional leadership. Interpretations related to hierarchy or upward communication should therefore be treated as contextual inference rather than as a direct empirical finding.
More broadly, the findings are context-specific and should not be generalized mechanically to all LMIC settings. Their value lies more in analytical transferability than statistical generalization. Finally, although the design allowed for triangulation between interview material and descriptive survey trends, it did not include classroom observation, student perspectives, or administrative decision-makers. Future work incorporating these perspectives would provide a more complete account of how GenAI readiness is negotiated across the wider institutional ecosystem.
6 Conclusion
This study explored how faculty in Bangladeshi higher education perceive and experience readiness for generative artificial intelligence (GenAI), drawing primarily on 17 in-depth interviews and supplemented by a descriptive faculty survey (n = 39). The findings suggest a landscape in which individual experimentation is advancing more quickly than institutional preparation. Faculty are already engaging with GenAI in varied and sometimes creative ways, but this engagement is taking place within conditions of policy ambiguity, uneven infrastructure, limited formal support, and contested professional norms.
At the individual level, the study indicates that familiarity with GenAI tools does not necessarily translate into deeper pedagogical readiness. Participants described a clear distinction between being able to use GenAI and being able to evaluate its outputs critically, integrate it into teaching responsibly, or align it with disciplinary expectations. The survey patterns support this interpretation, with confidence in tool use appearing more common than confidence in pedagogical design or critical evaluation. Training emerged as a major perceived need, yet the interviews showed that training is not understood as a neutral solution by all faculty. While some participants wanted structured support, others were willing to engage only under discipline-sensitive conditions, and a smaller group viewed training itself as a problematic signal of institutional endorsement.
At the institutional level, readiness appears fragmented rather than absent. Governance is often enacted informally through inherited plagiarism norms, improvised disclosure expectations, and contested detection tools, rather than through clear and coordinated GenAI-specific policy. Infrastructure is similarly partial. Access often depends on personal devices, shared facilities, free-tier tools, or unofficial subscription pathways, which raises questions not only about capacity but also about equity, legitimacy, and sustainability. In this sense, the study suggests that infrastructural readiness in resource-constrained contexts should be understood not only in terms of devices and connectivity, but also in terms of access integrity.
Faculty perceptions of GenAI were consistently ambivalent. Participants described the technology as useful for efficiency, support, and idea generation, while also associating it with risks to academic integrity, student effort, writing development, employability, and broader moral responsibility. These concerns were often embedded in culturally grounded understandings of diligence, professional accountability, and the social role of higher education. The findings therefore suggest that readiness for GenAI integration is shaped not only by technical knowledge or institutional policy, but also by local interpretive frameworks that influence whether GenAI is seen as a resource, a risk, or both at once.
Rather than offering a representative account of all Bangladeshi higher education institutions, this study provides an exploratory, context-sensitive understanding of how readiness is being interpreted and negotiated by faculty within a resource-constrained environment. It offers four main insights. First, resistance may be directed not only at GenAI tools themselves, but also at training and capacity-building initiatives. Second, access integrity appears to be an important but underexamined dimension of infrastructural readiness. Third, cultural and moral narratives are central to how faculty interpret readiness, not peripheral to it. Fourth, institutional variation is layered and cannot be reduced to a simple public–private distinction.
For higher education institutions in Bangladesh and similar LMIC settings, the findings point to three practical priorities. First, professional development should move beyond tool demonstrations to include critical evaluation, pedagogical redesign, and ethical reasoning. Second, governance frameworks should provide clear disclosure expectations and workable guidance for assessment without depending narrowly on detection technologies. Third, infrastructure planning should prioritize equitable and sustainable access, including legitimate licensing, secure access pathways, and closer integration with institutional systems.
Future research could extend this work by incorporating student, administrator, and policymaker perspectives, and by examining how faculty practices and institutional responses evolve over time. Comparative work across disciplines, institutional types, and national contexts would also help clarify which aspects of readiness are locally specific and which may be more broadly transferable.
Overall, the Bangladeshi case suggests that GenAI readiness in higher education is best understood as uneven, negotiated, and aspirational. Faculty are already making sense of the technology and constructing local norms around its use, but institutional policy, support, and infrastructure have not yet caught up. Responding effectively will require approaches that are not only technically informed, but also institutionally realistic and culturally grounded.
Ethics Statement
This study involved non-interventional social science research with adult participants. Formal approval from an institutional research ethics committee was not required under applicable national regulations and institutional practice in Bangladesh. The study was conducted in accordance with recognized ethical principles for research involving human participants. All participants received information about the purpose of the study, the voluntary nature of participation, and confidentiality protections prior to data collection. Written informed consent was obtained for in-person interviews, and verbal informed consent was obtained and recorded for online interviews. Data were anonymized to protect participant confidentiality.
Data Availability Statement
Anonymized survey data and selected interview excerpts are available from the corresponding author upon reasonable request. Full interview transcripts are not publicly available due to confidentiality and privacy considerations.
Funding
This research received no external funding.
Conflicts of Interest
The authors declare no conflict of interest.
Acknowledgments
The authors thank the faculty members from public and private universities in Bangladesh who participated in this study. Their openness and thoughtful reflections made this research possible. The authors also used ChatGPT (OpenAI) during manuscript preparation solely for language editing and clarity improvement. All content was reviewed, revised, and validated by the authors, who take full responsibility for the final manuscript.
Author Contributions
Wahid bin Ahsan: Conceptualization, methodology, formal analysis, investigation, data interpretation, writing (original draft), writing (review and editing), supervision, project administration, visualization.
Md. Osman Gani: Investigation, data curation, formal analysis, data interpretation, writing (review and editing).
Md. Shamsul Arefin: Investigation, data curation, formal analysis, data interpretation, writing (review and editing).
Samia Bari: Data curation, formal analysis, validation, data interpretation, writing (review and editing).
Tanim Ahamed: Data curation, formal analysis, validation, data interpretation, writing (review and editing).
Md. Shahadat Hossain: Data curation, formal analysis, validation, data interpretation, writing (review and editing).
M. H. Taohid: Data curation, formal analysis, validation, data interpretation, writing (review and editing).
All authors reviewed the manuscript, approved the final version, and agreed to be accountable for the work.
References
Al-kumaim, N. H., Hassan, S. H., Al-shami, S. A., & Alhazmi, A. K. (2025). Exploring Generative AI Usage Patterns in Universities: Analysis and Guidelines for Sustainable Practices. International Journal of Technology in Education, 8(2), 332–361. https://doi.org/10.46328/ijte.1045
Alli, A. A., Magombe, Y., Lwembawo, I., & Kasadha, J. (2025). A Policy Framework for the Use of Generative Artificial Intelligence in Higher Education Institutions. 2025 IST-Africa Conference (IST-Africa), 1–11. https://doi.org/10.23919/IST-Africa67297.2025.11060512
Alsharefeen, R., & Al Sayari, N. (2025). Examining academic integrity policy and practice in the era of AI: a case study of faculty perspectives. Frontiers in Education, 10. https://doi.org/10.3389/feduc.2025.1621743
An, Y., Yu, J. H., & James, S. (2025). Investigating the higher education institutions’ guidelines and policies regarding the use of generative AI in teaching, learning, research, and administration. International Journal of Educational Technology in Higher Education, 22(1), 10. https://doi.org/10.1186/s41239-025-00507-3
Chan, C. K. Y., & Hu, W. (2023). Students’ voices on generative AI: perceptions, benefits, and challenges in higher education. International Journal of Educational Technology in Higher Education, 20(1), 43. https://doi.org/10.1186/s41239-023-00411-8
Collie, R. J., & Martin, A. J. (2024). Teachers’ motivation and engagement to harness generative AI for teaching and learning: The role of contextual, occupational, and background factors. Computers and Education: Artificial Intelligence, 6, 100224. https://doi.org/10.1016/j.caeai.2024.100224
Gill, A. S., & Germann, S. (2022). Conceptual and normative approaches to AI governance for a global digital ecosystem supportive of the UN Sustainable Development Goals (SDGs). AI and Ethics, 2(2), 293–301. https://doi.org/10.1007/s43681-021-00058-z
Jin, Y., Yan, L., Echeverria, V., Gašević, D., & Martinez-Maldonado, R. (2024). Generative AI in Higher Education: A Global Perspective of Institutional Adoption Policies and Guidelines. Computers and Education: Artificial Intelligence. https://doi.org/10.48550/arXiv.2405.11800
Kohnke, L., & Ulla, M. B. (2024). Embracing generative artificial intelligence: The perspectives of English instructors in Thai higher education institutions. Knowledge Management & E-Learning: An International Journal, 653–670. https://doi.org/10.34105/j.kmel.2024.16.030
Krause, S., Panchal, B. H., & Ubhe, N. (2025). Evolution of Learning: Assessing the Transformative Impact of Generative AI on Higher Education. Frontiers of Digital Education, 2(2), 21. https://doi.org/10.1007/s44366-025-0058-7
Kumar, D. (2025). Impact of Infrastructure and Digital Literacy on Cloud-AI EdTech Adoption in Rural India. Edumania-An International Multidisciplinary Journal, 3(2), 258–268. https://doi.org/10.59231/edumania/9131
Li, M., Xie, Q., Enkhtur, A., Meng, S., Chen, L., Yamamoto, B. A., Cheng, F., & Murakami, M. (2025). A Framework for Developing University Policies on Generative AI Governance: A Cross-national Comparative Study. https://arxiv.org/abs/2504.02636
Mannuru, N. R., Shahriar, S., Teel, Z. A., Wang, T., Lund, B. D., Tijani, S., Pohboon, C. O., Agbaji, D., Alhassan, J., Galley, J., Kousari, R., Ogbadu-Oladapo, L., Saurav, S. K., Srivastava, A., Tummuru, S. P., Uppala, S., & Vaidya, P. (2025). Artificial intelligence in developing countries: The impact of generative artificial intelligence (AI) technologies for development. Information Development, 41(3), 1036–1054. https://doi.org/10.1177/02666669231200628
Maxwell, D., Oyarzun, B., Kim, S., & Bong, J. Y. (2025). Generative AI in Higher Education: Demographic Differences in Student Perceived Readiness, Benefits, and Challenges. TechTrends. https://doi.org/10.1007/s11528-025-01109-6
Modiba, F. S., Van den Berg, A., & Mago, S. (2025). Opportunities and challenges of generative artificial intelligence supporting research in African classrooms. South African Journal of Higher Education, 38(3). https://doi.org/10.20853/39-3-6272
Mutelo, I. (2025). Understanding the Generative Artificial Intelligence Revolution in Zambian Higher Education Research: Adoption, Challenges, and Strategies for Responsible Integration. International Journal of Research and Innovation in Social Science, IX(IIIS), 5731–5737. https://doi.org/10.47772/IJRISS.2025.903SEDU0416
Nevárez Montes, J., & Elizondo-Garcia, J. (2025). Faculty acceptance and use of generative artificial intelligence in their practice. Frontiers in Education, 10. https://doi.org/10.3389/feduc.2025.1427450
Nikolic, S., Wentworth, I., Sheridan, L., Moss, S., Duursma, E., Jones, R. A., Ros, M., & Middleton, R. (2024). A systematic literature review of attitudes, intentions and behaviours of teaching academics pertaining to AI and generative AI (GenAI) in higher education: An analysis of GenAI adoption using the UTAUT framework. Australasian Journal of Educational Technology. https://doi.org/10.14742/ajet.9643
Pramjeeth, S., & Ramgovind, P. (2024). Generative Artificial Intelligence (AI) Tools in Higher Education: A Moral Compass for the Future? African Journal of Inter/Multidisciplinary Studies, 6(1), 1–13. https://doi.org/10.51415/ajims.v6i1.1560
Shata, A., & Hartley, K. (2025). Artificial intelligence and communication technologies in academia: faculty perceptions and the adoption of generative AI. International Journal of Educational Technology in Higher Education, 22(1), 14. https://doi.org/10.1186/s41239-025-00511-7
UNESCO. (2023). Guidance for generative AI in education and research. UNESCO. https://doi.org/10.54675/EWZM9535
Valdivieso, T., & González, O. (2025). Generative AI Tools in Salvadoran Higher Education: Balancing Equity, Ethics, and Knowledge Management in the Global South. Education Sciences, 15(2), 214. https://doi.org/10.3390/educsci15020214
Wang, H., Dang, A., Wu, Z., & Mac, S. (2024). Generative AI in higher education: Seeing ChatGPT through universities’ policies, resources, and guidelines. Computers and Education: Artificial Intelligence, 7, 100326. https://doi.org/10.1016/j.caeai.2024.100326
Yusuf, A., Pervin, N., & Román-González, M. (2024). Generative AI and the future of higher education: a threat to academic integrity or reformation? Evidence from multicultural perspectives. International Journal of Educational Technology in Higher Education, 21(1), 21. https://doi.org/10.1186/s41239-024-00453-6