Edited by: Davide Girardelli, University of Gothenburg, Sweden
Reviewed by: Rashmi Kodikal, Graphic Era University, India
Gila Kurtz, Holon Institute of Technology, HIT, Israel
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Research on Generative Artificial Intelligence (GAI) in higher education primarily focuses on faculty use and experiences, with limited attention given to why some abstain from using it. Drawing from Innovation Resistance Theory, this study aims to address this gap by exploring the perceptions of both faculty users and non-users of GAI, identifying the reasons and concerns why they avoid GAI. A survey of 294 full-time higher education faculty from two mid-size U.S. public universities was conducted. Using qualitative and quantitative analysis, results show that over one-third of the faculty members opted out of using GAI for five primary reasons: not ready/not now, no perceived value, identity in tension, threat to human intelligence, and future fears and present risks. While both groups expressed concerns about academic dishonesty, non-users associate GAI with broader negative societal consequences, whereas users viewed it as related to innovation and potential benefits. For non-users, top concerns included a perceived lack of originality and accountability, while users were primarily concerned with accuracy. Surprisingly, general comfort with technology emerged as a significant predictor of non-user faculty’s behavioral intention to use GAI. This research contributes to understanding faculty resistance to GAI, emphasizing the need to balance its benefits with drawbacks in higher education.
As Generative AI (GAI) becomes more widespread, it has the potential to transform the landscape of higher education, bringing significant benefits and challenges that warrant careful consideration (
Past research on GAI in higher education has been largely exploratory, with a primary focus on students’ perceptions (
Innovation resistance theory (IRT) explains why individuals resist adopting new innovations (
Understanding faculty concerns and the reasons behind their hesitations is crucial for overcoming barriers to AI integration in academia, ensuring its ethical implementation that supports, rather than replaces, human creativity and critical thinking, while also preparing students for an AI-driven workforce. This research aims to explore and compare GAI perceptions and concerns among faculty non-users and users of AI, to understand the reasons why faculty abstain or do not use GAI, and to identify the factors affecting their decision to adopt. This will offer valuable insights that will help create a balanced, responsible approach to AI adoption in higher education, ensuring its benefits are maximized while minimizing potential risks. Thus, the following are proposed:
An online survey was designed using Qualtrics and administered to full-time higher education faculty members recruited from two mid-size public U. S. universities, one on the East Coast and one in the Southwest. The sample (
Faculty emails were compiled from an online directory, then narrowed to social sciences and humanities. Using Qualtrics, we sent email invitations to participate in the study along with a survey link. Upon clicking, participants first completed a consent form, followed by a filter question if they use GAI. Non-users were asked about their perceptions, concerns, reasons for non-use, behavioral intentions, and technology comfort, followed by demographics. Users answered the same questions, plus reported current GAI uses and impressions. To ensure data quality, attention-check questions were included to detect inattention or rushed responses. Data collection was conducted in 2024.
Thematic analysis was employed to identify, organize, and interpret patterns of meaning within the data. The process began with close reading to become familiar with the content and taking notes of initial observations. Using an inductive approach, the data was re-examined closely and assigned initial codes, grouping relevant parts together based on emerging concepts. Next, by bringing together the codes and fragmented ideas that link data together, analyzing data for patterns of similarity or contrast, and synthesizing codes into potential themes. Finally, the themes were refined and modified to reflect the data and patterns of shared meaning.
All the theoretical constructs used existing measures, adapted to fit the AI context and measured using multi-item scales validated in previous research.
GAI concerns among faculty members.
| Type of concerns | AI usage | Concern level | |||
|---|---|---|---|---|---|
| Major concern % | Moderate concern % | Minor concern % | Not a concern % | ||
| Plagiarism | Users | 54.9 | 25.6 | 14.4 | 5.1 |
| Non-users | 68.7 | 19.2 | 10.1 | 2.0 | |
| Copyright | Users | 47.2 | 33.3 | 12.8 | 6.7 |
| Non-users | 61.6 | 20.2 | 11.1 | 7.1 | |
| Accountability | Users | 54.4 | 28.2 | 13.8 | 3.6 |
| Non-users | 66.7 | 24.2 | 6.1 | 3.0 | |
| Originality | Users | 43.1 | 31.8 | 19.5 | 5.6 |
| Non-users | 72.7 | 20.2 | 4.0 | 3.0 | |
| Dependability | Users | 43.1 | 33.8 | 16.4 | 6.7 |
| Non-users | 41.4 | 37.4 | 14.1 | 7.1 | |
| Credibility | Users | 52.8 | 31.3 | 11.8 | 4.1 |
| Non-users | 54.5 | 32.3 | 10.1 | 3.0 | |
| Accuracy | Users | 61.0 | 27.7 | 9.2 | 2.1 |
| Non-users | 57.6 | 30.3 | 8.1 | 4.0 | |
| Privacy | Users | 42.6 | 28.2 | 21.5 | 7.7 |
| Non-users | 43.4 | 37.4 | 13.1 | 6.1 | |
| Safety | Users | 26.7 | 34.9 | 26.2 | 12.3 |
| Non-users | 32.3 | 44.4 | 14.1 | 9.1 | |
| Take your job | Users | 7.7 | 13.8 | 37.9 | 40.5 |
| Non-users | 14.1 | 19.2 | 29.3 | 37.4 | |
A total of 294 higher-education faculty completed the survey; 33.6% reported not using GAI. Faculty roles included Full Professors (24.5%), Associate Professors (23%), Assistant Professors (19%), Faculty in residence (7.5%), Adjunct Faculty (6.5%), and Visiting Faculty (6.5%). Most (82%) held no administrative roles. Experience levels were: 35% (20 + years), 21% (9–14 years), 20% (4–8 years), and 15% (15–20 years). Participants identified as Female (45%), Male (43%), and prefer not to say (8%). Participants identified as Caucasian (69%), African American (7%), Asian/Pacific Islander (7%), Hispanic (6%). Their age ranged from 41–50 (25%), 51–60 (23%), 31–40 (21%), and 61–70 (15%).
Thematic analysis highlighted five key themes/barriers:
A qualitative analysis of faculty members’ perceptions of GAI was conducted based on their brief descriptions of the GAI technology. Non-users associate GAI with academic dishonesty with concerns centered around cheating, plagiarism, copyright violations, inaccuracy, unethical, cheap shortcut, invasion of privacy, lacking deep third-level thought, causing “intellectual laziness.” Many express fears about the future for its possible negative societal impact, describing it with terms like the “Terminator, Skynet, Pandora’s box, dystopian future, uncharted, I-robot,” warning it could cause trouble or lead to significant complications, like perceived “loss of humanity,” dismissing it as “garbage.” However, some faculty acknowledge its powerful potential, describing it as helpful, smart, fast, and full of possibilities. While few faculty adopt a more balanced perspective, viewing it as a “double-edged sword - opportunity, yet scary, good and bad, strange but awesome.”
For faculty users, they perceive it as “an evolution, advanced technology or advanced Google” that has the potential to bring about “revolutionary change.” Many see significant benefits, particularly in its role as an “assistant” where it is praised for being “efficient, helpful, convenient, saves time, good for brainstorming, and a powerful tool” for boosting 3 Ps—productivity, prediction, and problem-solving. Participants successfully identified and described GAI as “LLMs, machine learning, and language model.” However, some faculty express concerns about its experimental nature, especially its unreliability and unpredictability, questioning its trustworthiness, asking questions “What if….—what is true?” This uncertainty about its future impact leaves them uneasy, viewing it as inherently risky. They share the same academic dishonesty concerns, in addition to “hallucinations, biased algorithms, lack of regulation, and reduced agency.”
As shown in
For H1, an independent t-test was conducted and results showed that GAI concerns were higher among faculty users (
For H2, a multiple regression found that the model was significant, [
The study findings address the gap in the literature by providing new empirical evidence into faculty perspective, moving beyond surface-level concerns to deeply explore the underlying reasons or barriers for opting out of GAI. The five themes/barriers align with the Innovation Resistance Theory (
Regarding faculty perceptions, users tend to emphasize the potential benefits of GAI, while non-users adopt a more cautious and pessimistic perspective. These findings align with
Faculty expressed concern about academic dishonesty, consistent with past literature (
Quantitative analysis revealed that non-users reported lower levels of general comfort with technology, which negatively influenced their intention to use GAI. Their reluctance appears rooted in a broader discomfort with new technologies that fall outside their comfort zone, potentially contributing to limited adoption. Although concerns about GAI were marginally insignificant (
This research offers valuable insights into the barriers driving faculty resistance to GAI in higher education, contributing to the literature on technology adoption/resistance. These findings challenge existing technology acceptance models by highlighting that faculty resistance is not solely a function of perceived usefulness or ease of use, but is also rooted in deeper identity-based, value/belief system, and ethical concerns. This suggests that current models must expand to account for the emotional and value-driven reasons, not just practical ones.
The findings inform the development of institutional policies that address faculty concerns through training, support, and clear guidelines. They also encourage ongoing academic dialog to balance GAI’s benefits with ethical implications, ensuring its use enhances rather than undermines human creativity and critical thinking, and prepares students for an AI-integrated future while preserving academic values. However, a key limitation of this study is focusing on two U.S. universities within the humanities and social sciences. Thus, findings may not be generalized to STEM disciplines or to institutions in international contexts. Future research can track changes over time and focus on disciplinary differences, especially in STEM.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
The studies involving humans were approved by UNLV Institutional Review Board. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.
AS: Investigation, Writing – review & editing, Formal analysis, Methodology, Writing – original draft, Data curation, Supervision, Conceptualization, Project administration.
The author(s) declare that financial support was received for the research and/or publication of this article. The publication fees for this article were supported by the UNLV University Libraries Open Article Fund..
The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The author declares that no Gen AI was used in the creation of this manuscript.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.