The rapid expansion of artificial intelligence (AI), particularly generative AI, in higher education is transforming teaching, learning, and assessment practices, while simultaneously raising ethical concerns related to academic integrity, authorship, and critical thinking. Understanding teachers’ ethical perceptions is essential for responsible AI integration in academic environments.
A mixed-methods, cross-sectional study was conducted, combining quantitative descriptive and correlational analyses with qualitative content analysis. Data were collected through an online questionnaire ad-ministered to 119 professors from a Portuguese higher education institution in the Alto Alentejo region. Quantitative data were analyzed using descriptive statistics and non-parametric tests, and open-ended responses were examined using thematic content analysis.
The findings reveal widespread use of AI tools among professors, mainly for pedagogical content creation and document summarization. The most frequently identified ethical challenges were authorship and plagiarism, academic integrity, misinformation, and lack of regulation. Younger and female professors reported higher levels of self-reported ethical awareness indicators, while no significant association was found between frequency of AI use and ethical sensitivity. Qualitative analysis highlighted concerns related to critical thinking, assessment fairness, and the preservation of human-centered pedagogy.
This study examines higher education professors’ ethical awareness regarding the use of artificial intelligence, focusing on perceived ethical challenges, knowledge of ethical principles, and experiences with ethical dilemmas.
Artificial Intelligence (AI) remains a fluid and evolving concept, which makes it difficult to establish a universally accepted definition. This conceptual openness reflects the complexity and the highly interdisciplinary nature of AI, while also revealing the constant technological transformations that expand what is considered “intelligent.” In higher education, such conceptual ambiguity contrasts with the rapid adoption of AI systems—particularly generative AI tools—that are reshaping teaching, learning, assessment, and academic administration.
Recent literature shows a sharp increase in publications on AI in education since 2022, corresponding to the broad public adoption of generative AI. Recent systematic reviews have highlighted both the pedagogical potential of AI and the ethical challenges it raises in higher education contexts (
Research has documented significant benefits related to the integration of AI into higher education. Adaptive learning systems can adjust instructional content and pace to students’ individual needs, leading to more personalized and potentially more equitable learning experiences (
AI may further contribute to increased accessibility and inclusiveness, by supporting different learning profiles, helping to reduce language barriers, and expanding access to high-quality educational resources regardless of socioeconomic or geographic constraints (
Nevertheless, alongside these opportunities come extensive ethical, pedagogical, and institutional concerns. Current studies warn of risks associated with academic integrity, excessive technological dependence, biased datasets, inequitable access, and insufficient critical AI literacy among users (
In response to these challenges, scholars emphasize that AI implementation should not be approached merely as a technical upgrade. It requires responsible and strategic governance, teacher training, clear institutional policies, and frameworks that ensure transparent, ethical, and purposeful use of AI in learning environments (
Given this complex and evolving landscape, the present study adopts a broad under-standing of AI — acknowledging its diversity, pedagogical possibilities, and associated dilemmas — and provides a critical perspective on its role in higher education. The objective is to contribute to an informed and balanced interpretation of AI in universities today: one that recognizes its potential while ensuring that ethical principles, academic values, and student development remain at the forefront.
In this study, ethical awareness is adopted as the overarching analytical construct. Ethical awareness is understood as a multidimensional concept encompassing individuals’ perceptions of ethical implications, their self-reported knowledge of ethical principles, sensitivity to potential ethical dilemmas, and their capacity to recognize ethical risks associated with AI use. Terms such as ethical concerns, ethical perceptions, and ethical sensitivity are therefore used to refer to specific dimensions or manifestations of ethical awareness, rather than distinct or competing constructs.
This study was developed with the aim of identifying the main ethical concerns associated with the use of Artificial Intelligence in academic environments, according to the perspectives of higher education professors at a Portuguese institution, in the Alto Alentejo region.
The questionnaire was designed to capture multiple dimensions of ethical awareness, including perceived ethical implications, familiarity with ethical principles, and experiences of ethical dilemmas related to AI use.
Hypothesis: Based on the objectives and theoretical framework of this study, the following hypotheses were formulated:
H1. Younger and female professors report higher levels of self-reported ethical awareness indicators regarding the use of Artificial Intelligence in academic contexts.
H2. The frequency of AI use is positively associated with ethical awareness and experience of ethical dilemmas.
H3. Professors who report higher levels of ethical awareness are more likely to identify authorship, plagiarism, and academic integrity as central ethical challenges in the use of artificial intelligence.
H4. Professors who report higher levels of self-reported ethical awareness indicatorsare more likely to identify teacher training as a key factor in promoting ethical AI use in academic contexts.
H5. Professors who report higher levels of self-reported ethical awareness indicators are more likely to emphasize the importance of institutional and legal regulation in ensuring ethical AI practices in higher education.
These hypotheses guided the quantitative component of the study, particularly the correlational analyses. The qualitative content analysis was conducted as a complementary and exploratory approach, aimed at deepening the interpretation of professors’ ethical concerns and perceptions regarding AI use. Hypotheses 3–5 were formulated to examine relationships between ethical awareness and professors’ perceptions of specific ethical challenges and institutional responses to AI use.
The hypotheses were formulated to guide the analytical exploration of associations between variables rather than to test causal or strongly confirmatory propositions. Accordingly, results are interpreted as indicative patterns rather than definitive tests of theory.
This study adopted a mixed-methods, cross-sectional design, integrating a quantitative descriptive and correlational approach with qualitative content analysis. Quantitative and qualitative data were collected concurrently and analyzed separately, with integration occurring at the interpretation stage. Used an online questionnaire created in Google® Forms, applied to professors from a Portuguese higher education institution. Data were initially stored on Google Forms’ secure cloud servers during collection and subsequently downloaded to the researchers’ password-protected personal computers, where confidentiality and anonymity were ensured in compliance with the General Data Protection Regulation (GDPR). Data collection took place from January to May 2025. The questionnaire included
The questionnaire was an
In the Portuguese higher education context, the term “professor” is used to refer to all members of the academic teaching staff, including lecturers and assistant professors, and does not exclusively denote the highest academic rank.
The survey was distributed to the entire target population of the institution (
Participation was voluntary, and no follow-up data were collected from non-respondents. Non-response may be related to factors such as time constraints, varying levels of interest in the topic, or differing familiarity with AI tools.
Ethical considerations were always respected, and procedures followed the Declaration of Helsinki. Data confidentiality was guaranteed, with information stored on researchers’ personal computers secured by access codes. Although no direct identifying information (such as names or email addresses) was collected, we acknowledge that, in a small institutional context, certain combinations of demographic variables (e.g., gender, field, or position) could potentially allow indirect identification. To mitigate this risk, all data were anonymized prior to analysis, and results are presented only in aggregated form to ensure participants’ confidentiality in compliance with the GDPR and the Declaration of Helsinki. The study was approved by the Ethics Committee of the institution (Ethics Opinion No. SC/2024/34, 14 November 2024). The questionnaire included self-report items designed for this study, such as: sex; age; academic qualifications; field of expertise; teaching experience in higher education; frequency of AI use; reasons for using AI; most frequently used AI tools; ethical implications of AI use; ethical dilemmas encountered; knowledge of ethical principles in AI use; ethical issues associated with AI; and what participants considered most important to promote ethical conduct in AI use.
Before data collection, a pilot test was conducted with 10 professors. After completing the survey, no substantive issues were identified by the pilot participants, and the initial version was therefore maintained, as it was considered appropriate, clear, and easy to complete.
Quantitative data were analyzed using IBM SPSS Statistics, version 27. Descriptive statistics (frequencies, percentages, means, and standard deviations) were used to characterize the sample and summarize patterns of AI use and ethical perceptions. Given the ordinal nature of most variables and the non-normal distribution of the data, non-parametric statistical tests were applied. Spearman’s rank-order correlation was used to examine associations between sociodemographic variables (age, gender, teaching experience), frequency of AI use, and dimensions of ethical awareness. Group comparisons were conducted using the Mann–Whitney U test and the Kruskal–Wallis test, as appropriate, to explore differences across categories such as academic degree and type of AI access. Statistical significance was set at
In addition to significance testing, effect sizes were calculated to assess the magnitude of associations. Spearman’s rank-order correlation coefficients (r
Given the ordinal nature of the variables and deviations from normality, non-parametric tests were selected. Although multiple statistical tests were conducted, no formal correction for multiple comparisons was applied due to the exploratory nature of the study; therefore, findings should be interpreted with caution regarding potential inflation of Type I error.
The study also employed content analysis to interpret professors’ responses to an open-ended question about their ethical concerns regarding AI use. This qualitative technique enabled a systematic exploration of expressed perceptions, organizing them into analytical categories based on defined criteria.
The analysis unfolded in several stages, beginning with familiarization with the empirical corpus, followed by the progressive construction of a coding system grounded in
Coding involved converting raw data into meaningful units representing ex-pressed content. Categorization grouped statements with common characteristics into thematic categories. Text was segmented into units of meaning—ranging from single words to broader excerpts—enabling more precise and theoretically informed analysis (
Following
Pre-analysis—floating reading and organization of collected material to systematize initial impressions and outline an analytical plan.
Exploration—coding and categorization of data according to defined criteria.
Results treatment and interpretation—inference and analysis of categorized units based on identified patterns.
The categories were refined throughout the analysis, accommodating significant new units, which enhanced interpretative accuracy (
According to
The open-ended responses were analyzed using qualitative content analysis. Two researchers independently coded the data following an inductive–deductive approach informed by the study objectives and the literature. Initial coding was conducted independently, after which the coders met to compare coding decisions and discuss discrepancies. Disagreements were resolved through consensus, leading to the refinement of category definitions and indicators.
Given the exploratory nature of the study, formal intercoder reliability statistics (e.g., Cohen’s kappa) were not calculated. Instead, credibility was supported through independent coding, iterative discussion, and the maintenance of an audit trail documenting coding decisions, category revisions, and inclusion criteria.
The sample included 119 participants, representing 46% of the total number of professors at a Portuguese higher education institution. The majority were between 41 and 60 years old (74.8%), held a Ph.D. (58.8%), had training in the social sciences and humanities (47.9%), taught at a school of technology, management, and design (47.1%), and had been teaching for more than 20 years (40.3%) (
Sociodemographic, academic, and professional characterization of the sample.
| Sociodemographic, academic, and professional characteristics | % | ||
|---|---|---|---|
| Gender | Male | 59 | 49.6 |
| Female | 59 | 49.6 | |
| No response | 1 | 0.8 | |
| Age | Up to 25 years | 0 | 0 |
| 25–30 years | 4 | 3.4 | |
| 31–40 years | 16 | 13.4 | |
| 41–50 years | 44 | 37 | |
| 51–60 years | 45 | 37.8 | |
| Over 60 years | 10 | 8.4 | |
| Academic qualification | Bachelor’s degree | 13 | 10.9 |
| Master’s degree | 30 | 25.2 | |
| Ph.D. | 70 | 58.3 | |
| Pós-Ph.D. | 6 | 5 | |
| Field of expertise | Exact and technological sciences | 25 | 21 |
| Social and human sciences | 57 | 47 | |
| Arts | 9 | 7.6 | |
| Health and life sciences | 22 | 18.5 | |
| Other | 6 | 5 | |
| Years of teaching in higher education | Less than 5 | 40 | 33.6 |
| 5–10 | 15 | 12.6 | |
| 11–20 | 16 | 13.4 | |
| More than 20 | 48 | 40.3 | |
Regarding the way in which AI is used, the majority reported using it sometimes (42.9%) and accessing it freely without a subscription (74.8%) (
Form of artificial intelligence use.
| Form of AI use | % | ||
|---|---|---|---|
| Use of AI | Does not use | 7 | 5.9 |
| Yes, but rarely | 33 | 27.7 | |
| Yes, sometimes | 51 | 42.9 | |
| Yes, very often | 20 | 16.8 | |
| Yes, every day | 8 | 6.7 | |
| AI tools subscription | Uses AI with paid subscription | 21 | 17.6 |
| Accesses AI tool without subscription | 89 | 74.8 | |
| Not applicable | 9 | 7.6 | |
As for the reasons for using AI, the most frequently mentioned were summarizing documents (42%), creating pedagogical content (38.7%), generating text on a specific subject (38.5%), and conducting more efficient online searches and/or navigation (36.1%) (
Main reasons for artificial intelligence (AI) use.
| Main reasons for AI use | % | ||
|---|---|---|---|
| Using AI for | Producing academic papers | 15 | 12.6 |
| Producing scientific articles | 11 | 9.2 | |
| Creating pedagogical content | 46 | 38.7 | |
| Summarizing documents | 50 | 42 | |
| Controlling smart electronic devices | 6 | 5 | |
| Summarizing or proofreading documents | 35 | 29.4 | |
| Applying or improving writing styles | 39 | 32.8 | |
| Generating text on a specific topic | 47 | 38.5 | |
| Generating or editing images | 26 | 21.8 | |
| Generating or editing videos | 7 | 5.9 | |
| Consulting specialized tutorials | 12 | 10.1 | |
| More efficient online search/navigation | 43 | 36.1 | |
| Administrative management support | 12 | 10.1 | |
| Planning classes or research/intervention projects | 32 | 26.9 | |
| Assessment and feedback | 18 | 15.1 | |
| Other uses | 24 | 20.2 | |
| Not applicable | 8 | 6.7 | |
The AI tools most frequently used by respondents were Chatbot—ChatGPT (83.2%), Chatbot—Microsoft Copilot (26.1%), and Chatbot—Google Gemini (21.8%) (
Artificial intelligence (AI) tools most frequently used.
| Used AI tools | |||
|---|---|---|---|
| % | |||
| AI tools | Chatbot—ChatGPT | 99 | 83.2 |
| Chatbot—Google Gemini | 26 | 21.8 | |
| Chatbot—Microsoft Copilot | 31 | 26.1 | |
| Chatbot—DeepSeek | 11 | 9.2 | |
| DeepL translator | 50 | 42 | |
| Marketing—EllieAI | 5 | 4.2 | |
| Productivity—MagicalAI | 1 | 0.8 | |
| Productivity—QuestAI | 2 | 1.7 | |
| Video—Filmora | 2 | 1.7 | |
| IBM watson | 1 | 0.8 | |
| Images—Midjourney | 3 | 2.5 | |
| Images—Dall-E | 8 | 6.7 | |
| Other | 30 | 25.2 | |
| Not applicable | 7 | 5.9 | |
With regard to ethical challenges and AI use, most respondents reported that they always consider the ethical implications of AI use (62.9%), when excluding “not applicable” responses, had never faced an ethical dilemma (78.2%), and felt moderately informed about the ethical principles of AI use (51.3%) (
Ethical perception in the use of artificial intelligence.
| Ethical perception | % | ||
|---|---|---|---|
| Considers the ethical implications of AI use | Always | 82 | 62.9 |
| Frequently | 20 | 16.8 | |
| Occasionally | 3 | 2.5 | |
| Never | 3 | 2.5 | |
| Not applicable | 11 | 9.2 | |
| Feels informed about ethical principles in AI use | Yes, clearly | 27 | 22 |
| Yes, moderately | 61 | 51.3 | |
| Yes, vaguely | 21 | 17.6 | |
| No | 10 | 8.4 | |
| Has faced any ethical dilemma in AI use | Yes | 26 | 21.8 |
| No | 93 | 78.2 | |
Respondents who reported having encountered an ethical dilemma mentioned situations such as: “Creating documents as if they were my own work when they are not. I wonder whether I should consider them finished and use them,” “Comical situations in which students present excellent texts but do not understand what is written—how should I evaluate them?” “In a piece of work I do, I am not sure which source to reference. The work is not mine, but I do not know who the author is. Should I use it as if it were my own?” “To what extent is the authorship of the work really mine?” and “In a text, I do not know the full accuracy of the information, so how much can I trust it?”
The most frequently mentioned ethical issues were the creation of content without effort, depersonalized but presented as one’s own authorship (76.6%); the risk of generating misinformation in a highly effective manner to manipulate or deceive (75.5%); and the lack of regulation and legal responsibility in the use of AI (68.1%) (
Ethical issues associated with artificial intelligence (AI) use.
| Ethical issues | % | ||
|---|---|---|---|
| Main ethical issues associated with AI use | Decrease in privacy and data security, threatening civil liberty | 52 | 43.7 |
| Discriminatory decisions made by algorithms that may negatively affect minority or vulnerable groups | 40 | 33.6 | |
| Difficulty assigning responsibility for decision proposals in medical diagnoses or legal opinions | 27 | 22.7 | |
| Difficulty balancing the economic benefits of AI with the social impacts of reduced job availability due to automation | 34 | 28.6 | |
| Difficulty assigning moral responsibility in decision-making that affects human life | 45 | 37.8 | |
| Risk of creating highly effective disinformation, manipulating or deceiving to achieve goals | 90 | 75.5 | |
| Possibility of “smart war” with autonomous weapons beyond human control | 24 | 20.2 | |
| Lack of regulation and legal responsibility in AI use | 81 | 68.1 | |
| Greater human vulnerability due to distorted perception of reality from excessive use | 62 | 52.1 | |
| Uncertain fate of humanity in the presence of “machines” with superior intelligence | 19 | 16 | |
| Creation of effortless, depersonalized content presented as one’s own authorship | 91 | 76.6 | |
| Other | 12 | 10.1 | |
| None | 4 | 3.4 | |
The majority also reported that AI should be regulated to ensure ethical practices (77.3%) and considered teacher training on this subject as the most important factor in promoting ethical conduct in AI use (36.1%) (
Most important interventions to promote ethical conduct in artificial intelligence (AI) use in higher education.
| Ethical issues | % | ||
|---|---|---|---|
| What is considered most important to promote ethical conduct in AI use in higher education | Teacher training | 43 | 36.1 |
| Institutional policies and guidelines | 34 | 28.6 | |
| Legal framework | 26 | 21.8 | |
| Investment in safer and more transparent technologies | 14 | 11 | |
| Other | 2 | 1.7 | |
Spearman’s correlation tests were applied to assess associations between age, gender, field of training, and teaching experience with three main variables: perceptions of AI’s ethical implications, degree of knowledge about ethical dilemmas, and previous experience with ethical dilemmas.
A weak but statistically significant positive correlation was observed between gender and reported degree of information about ethical dilemmas (ρ = 0.203,
Spearman’s correlations were used to evaluate whether the frequency of AI use was related to perceptions of ethical implications, experience with dilemmas, and the importance attributed to ethical conduct. No statistically significant correlations were observed (
No significant correlation was found between professional characteristics such as field of training and teaching experience and specific ethical concerns (
The Kruskal–Wallis test was used to compare groups with different academic degrees (bachelor’s, master’s, and Ph.D.) regarding frequency of AI use, degree of information about ethics, and experience with dilemmas. In all comparisons, the results were not significant (
The Mann–Whitney U test showed no significant difference between the groups defined by type of AI access (free or paid) in terms of the importance attributed to ethics (U = 1828.0,
The subdivision into categories and indicators is presented below (
Subcategories and indicators by category.
| Category | Subcategory | Indicators |
|---|---|---|
| Critical thinking, reflection, learning, and pedagogical mediation | Promotion of critical thinking | AI as a tool for questioning, reflection, and critical analysis |
| Pedagogical mediation with AI | Encouragement of argumentation and comparison of perspectives | |
| Reflection on learning | Reconfiguration of the teacher’s mediating role in AI-supported learning; Risk of reduced student autonomy and critical engagement | |
| Authorship, plagiarism, and academic integrity | Difficulties in defining authorship | Confusion between human authorship and AI-generated content |
| Plagiarism and originality | Unreferenced or undisclosed use of AI-generated material | |
| Transparency in AI use | Difficulty identifying the original source of AI-generated outputs; Challenges in attributing academic responsibility | |
| Assessment and fairness | Equity in assessment | Difficulty distinguishing between student-produced work and AI-generated output |
| Redefinition of assessment criteria | Limited effectiveness of traditional plagiarism detection tools; Need to redesign assessment methods to ensure fairness | |
| Truthfulness and quality of information | Reliability of content | Risk of inaccurate or fabricated information (AI hallucinations) |
| Verification capacity | Difficulty verifying sources and factual accuracy of AI-generated content; Persuasive but potentially misleading AI outputs | |
| Misuse of AI and lack of training | Use without critical awareness | Use of AI tools without understanding their limitations |
| Insufficient ethical and digital literacy | Need for training in ethical and responsible AI use; Lack of competencies to critically evaluate AI-generated outputs | |
| Ethical and institutional regulation | Standards and codes of conduct | Absence or insufficiency of institutional policies regulating AI use |
| Institutional responsibility | Need for clear institutional guidelines on acceptable AI practices; Institutional responsibility in monitoring ethical compliance | |
| Philosophical and ontological impacts | Nature of artificial intelligence | Questions regarding agency, creativity, and consciousness |
| Redefinition of the human | Ethical implications of delegating cognitive and creative tasks to machines | |
| Challenges to teaching and pedagogical practice | Curriculum adaptation | Increased pedagogical workload associated with AI integration |
| Workload and complexity | Uncertainty regarding the evolving role of the teacher |
The results of the distribution of respondents’ answers by different thematic categories are presented below, indicating the absolute value (
Categories identified in the content analysis of open responses (
| Category | Frequency ( |
Percentage (%) |
|---|---|---|
| Critical thinking, reflection, learning, and pedagogical mediation | 20 | 29.4 |
| Authorship, plagiarism, and academic integrity | 15 | 22.1 |
| Assessment and fairness | 9 | 13.2 |
| Truthfulness and quality of information | 8 | 11.8 |
| Misuse of AI and lack of training | 7 | 10.3 |
| Ethical and institutional regulation | 6 | 8.8 |
| Philosophical and ontological impacts | 3 | 4.4 |
| Challenges to teaching and pedagogical practice | 2 | 2.9 |
| Total | 68 | 100.0 |
The content analysis of professors’ open-ended responses revealed a predominance of concerns associated with critical thinking, reflection, and learning (29.4%), emphasizing the value of the teacher’s role in pedagogical mediation and in developing higher-order cognitive skills, even in a context of increasing AI presence. Authorship, plagiarism, and academic integrity appeared as the second most representative category (22.2%), reflecting anxieties about the originality of academic work and the need to reinforce ethical values in knowledge production. Other concerns were related to evaluation criteria and fairness (12.7%), the truthfulness and quality of AI-generated information (11.1%), the misuse of AI and lack of training (9.5%), ethical and institutional regulation (7.9%), deeper philosophical impacts (4%), and challenges to teaching practice (3.2%). These results reinforce that professors perceive AI not only as a technological tool but as a complex phenomenon with ethical, pedagogical, and social implications.
Small but statistically significant associations were observed between age/gender and selected self-reported ethical awareness indicators. Given the exploratory nature of the measures and the observed effect sizes, these associations should be interpreted with caution.
A total of 68 recording units were identified and categorized. Frequencies reported in
Methodologically, it should be noted that the unit of analysis was the unit of meaning (and not the number of participants), so some responses were coded into more than one category, reflecting their semantic richness and complexity.
The present study offers a comprehensive and updated view of higher education professors’ perceptions of the ethical challenges associated with Artificial Intelligence (AI), revealing results that both align with international findings and reflect the specific characteristics of the Portuguese context, particularly in the Alto Alentejo region. The results indicate that although AI is increasingly present in academic environments, ethical reflection does not always accompany technological adoption. Despite the majority of professors reporting never having faced ethical dilemmas involving AI, qualitative responses reveal deep concerns about authorship, plagiarism, academic fraud, and the erosion of critical thinking skills — suggesting that ethical dilemmas may exist but fail to be recognized as such. This distinction between implicit perception and explicit acknowledgment echoes the growing consensus in the literature that educators often consider AI-related ethical issues as technical or institutional matters rather than as personal moral responsibility (
Our data demonstrate that professors primarily associate ethical challenges with threats to academic integrity, particularly the possibility of students using AI-generated content without transparency. These concerns are consistent with studies reporting a rise in authenticity issues and concealment of AI-generated materials in academic work (
A notable finding of this study concerns ambivalence in professors’ perceptions. While they recognize AI’s capacity to support students in tasks such as writing, research, and accessibility, they simultaneously fear its potential to undermine the development of essential competencies, such as creativity, autonomy, and evaluation of information credibility. These tensions reflect international research that frames generative AI as a technology capable of both improving and weakening learning processes depending on how it is pedagogically integrated (
Institutional training emerged as the most demanded intervention among participants, confirming results from multiple countries in which educators acknowledge insufficient preparation to evaluate and integrate AI responsibly into teaching practices (
The study also showed that concerns were not equally distributed across the sample: younger and less experienced academics more often expressed critical perspectives on AI-related risks than older colleagues. Although this pattern must be interpreted with caution, similar generational differences have been reported internationally, where younger academic communities demonstrate higher exposure to debates on digital ethics and emerging technologies (
An interesting and somewhat paradoxical finding concerns the fact that a substantial proportion of respondents reported never having faced ethical dilemmas related to AI use, while simultaneously expressing strong concerns about issues such as authorship, assessment fairness, and institutional regulation. This apparent discrepancy may reflect differences in how respondents interpret the notion of an “ethical dilemma,” a normalization of AI use in everyday academic practices, or the absence of explicit institutional policies that would frame certain situations as ethically problematic. Rather than indicating a lack of ethical concern, this pattern may suggest implicit or unarticulated ethical tensions.
The qualitative dimension of this study also highlighted concerns regarding misinformation and the credibility of AI-generated responses, particularly in academic re-search contexts. Professors identified risks associated with the persuasive fluency of AI-generated output, which may appear accurate even when containing factual errors or fabricated references — a phenomenon widely documented in recent literature (
Taken together, these findings point to a necessary reconfiguration of the educational ecosystem. Participants called for clearer rules, updated assessment practices, inclusion of oral evaluation and in-person supervision, and design of authentic learning tasks less susceptible to automation. These suggestions are aligned with international proposals advocating diversification of assessment formats to minimize AI-enabled fraud and ensure pedagogical meaningfulness (
Finally, this study contributes to addressing a significant national gap: the scarcity of empirical research on the ethical implications of AI in Portuguese higher education. By providing localized evidence from the Alto Alentejo region, the study offers valuable support for informed decision-making, bridging a gap between global discourse and local reality. Continued research in this field is essential to monitor evolving professional practices, identify new dilemmas arising from technological development, and support the design of policies and training programs that effectively reconcile innovation with ethical responsibility. As institutions increasingly adopt generative AI, strengthening ethical awareness and institutional governance will be fundamental to safeguarding academic values and ensuring equitable and sustainable digital transformation.
This research makes an innovative contribution to the Portuguese context, where empirical studies on AI in higher education are still incipient. By mapping professors’ concerns and perceptions, it offers evidence that may inform the development of institutional policies, the design of training programs, and the adaptation of assessment strategies, promoting a more ethical and reflective approach to the use of AI in academic environments.
This study provides an innovative empirical contribution by being the first to map Portuguese professors’ ethical perceptions of AI, revealing generational and gender patterns and highlighting context-specific dilemmas in authorship and assessment. Based on these findings, training initiatives should prioritize the cultivation of ethical awareness, case-based reflection, and digital literacy aligned with the EU Artificial Intelligence Act (2024). Institutional policies should further encourage participatory dialogue and localized ethical frameworks that address the specific realities of Portuguese higher education.
The findings indicate a perceived need for teacher training and clearer institutional governance frameworks to address ethical challenges associated with AI use in higher education. However, this study does not evaluate the effectiveness of specific interventions. Accordingly, these recommendations should be understood as reflecting participants’ perceptions and the challenges identified, rather than evidence of intervention impact.
Nevertheless, the study also presents limitations, particularly regarding the size and non-probabilistic nature of the sample, which was restricted to a single higher education institution. Future studies should expand the scope of analysis to other national and international contexts, adopting longitudinal and comparative approaches to allow a deeper understanding of the evolution of professors’ perceptions and practices in the face of ongoing technological advances.
While the regional focus provides important contextual insight into a less-studied educational setting, it may limit the transferability of findings. Future studies should consider broader multi-institutional or comparative designs to enhance generalizability.
Ethical awareness was assessed using individual self-report indicators rather than a validated multidimensional scale, which limits the strength of trait-based or causal interpretations. Future research should aim to develop and validate comprehensive instruments to measure ethical awareness and sensitivity in the context of AI use in higher education.
This study was conducted within a single higher education institution using a cross-sectional design and voluntary participation. As such, the findings should be interpreted as context-specific and exploratory, and generalization to the broader Portuguese higher education system should be approached with caution.
Ultimately, Artificial Intelligence should not be seen as a threat but as an opportunity to reaffirm the central role of human mediation in education. For this to be possible, it is necessary to implement robust institutional policies, invest in continuous training, and foster reflective spaces that promote ethical awareness and responsible integration of AI in higher education.
The original contributions presented in this study are included in this article/supplementary material, further inquiries can be directed to the corresponding author.
The studies involving humans were approved by Ethics Committee of Polytechnic University of Portalegre. 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. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
AO: Writing – original draft, Writing – review & editing. MC: Writing – original draft, Writing – review & editing. OP: Writing – original draft, Writing – review & editing. BM: Writing – original draft, Writing – review & editing. CG: Writing – original draft, Writing – review & editing.
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
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.
Edited by:
Reviewed by: