AI in HE is a novel and constantly evolving issue. Research emphasizes the synergy between ethics and responsibility, as well as privacy and security, when using AI in academic contexts (Mora Naranjo et al., 2023). In this context, the need arises to generate specific policies and regulations at the corresponding institutional levels (Rodríguez-Argueta, 2020), as well as updates that integrate digital competencies and ethical awareness in study plans and programs (Vélez-Rivera et al., 2024).

The responses and reactions this technology has provoked among faculty members, specifically in their perceptions, evaluations, and positions, have been addressed in various studies. Specifically, the works by Humble and Mozelius (2022), McGrath et al. (2023), and Svaricek (2024) depict university professors who express skepticism about the benefits of AI for teaching, with concerns rooted in uncertainties about its impact on the educational context. The lack of knowledge about AI, particularly its practical applications and potential as a didactic resource (Sánchez, 2023), is contributing to faculty members feeling insecure and adopting a cautious stance, which has slowed its integration in specific areas of teaching, research, and/or university management.

On the other hand, University students are also reluctant to use AI tools due to social stigmatization and being accused of academic dishonesty. The academic environment has traditionally valued individual effort and achievement, and some students worry that using AI may be perceived as a lack of interest, convenience, lack of commitment, effort, or the authenticity of their work. Against this backdrop, experiences of integrating AI into HE curricula have begun to emerge (Gairín & Alguacil, 2024). The guidelines suggested by González Alcaide (2024) offer insights into its integration into teaching practices. The key seems to lie in knowledge and training for the effective, ethical, and responsible use of these tools (Mora Naranjo et al., 2023), building trust and encouraging their open use in the university environment, prioritizing ethical considerations when adopting AI technologies (Vélez-Rivera et al., 2024) to ensure equity, transparency, and data privacy, the goal of contributing to a more just and inclusive society.

From the second perspective, the existence of normative frameworks for AI use, research warns that in the current academic context, some HE institutions may still lack clear guidelines for the use of AI in academic work, aimed at both students and faculty members. The study by Seldon et al. (2020) highlights the impact of unregulated AI use on the academic community. Although the literature remains limited, the need for and importance of generating specific policies, regulations, and ethical norms for the responsible use of AI in educational curricula and programs are emphasized (Nguyen et al., 2023; Rodríguez-Argueta, 2020; Vélez et al., 2024), with the aim of leveraging the advantages that this tool can bring to academic and educational trajectories.

To tackle this challenge, The work of Chan (2023) can be used as a reference, which presents a political framework for ecological education in AI under three dimensions: pedagogical, governance, and operational. The author suggests exploring the integration of AI in university teaching through this ethical framework to enhance the understanding of its implications. In a similar vein, the study by Flores and García (2023, p. 45) advocates for the development of consensus-based public policy frameworks to regulate and raise awareness of the ethical use of AI in education. According to both authors:

This must involve the design and implementation of a Digital and AI Literacy Plan, considering that the main obstacle to the widespread adoption of AI tools is the lack of solid evidence of their effectiveness and impact on students' academic achievements. Therefore, it is necessary to improve governance, accessibility, and the reliability of AI.

From the third perspective -training in digital ethics-, several considerations should be taken into account. While research highlights the critical importance of ethics and responsibility in the application of AI (Mora Naranjo et al., 2023), are students and faculty truly trained in this regard? How do they use AI? Is it critical, ethical, and responsible use? In short, are they competent in digital ethics? The reviewed studies emphasize the need to strengthen faculty training plans in digital ethics (Marín & Tur, 2024), and even assess the impact that such plans are having on students (Mora Naranjo et al., 2023); are they genuinely fostering the responsible use of AI tools? At this point, the work of Vélez-Rivera et al. (2024) underscores the need to work with the academic community, faculty, and students, in the use of these tools, in alignment with safe, ethical, and responsible academic environments.

In the face of this challenge, research highlights the Importance of digital ethics training for university faculty members. González Alcaide (2024) presents a model of digital literacy for teachers, while Flores and García (2023, p. 45) speak of an "AI Literacy Plan which not only trains teachers in technical skills but also in ethical and philosophical debates". On the other hand, research emphasizes the need to prepare students to face the ethical challenges associated with AI. For example, Lopezosa et al. (2023) propose a core AI training model tailored to the educational field. It includes competencies, objectives, and essential study topics on AI, and is structured to fit into a broader curriculum, for undergraduate/graduate studies or specialized seminars. In a similar vein, González Alcaide (2024) develops a didactic unit aimed at students to promote appropriate use of AI as well as optimal performance of these tools. The proposal by Aler et al. (2024) also follows a similar direction. In brief, it focuses on educational strategies, ethical competencies, and resources necessary for the trustworthy implementation of AI across various disciplines of HE. Its goal is that the next generation of young people can contribute to the use of ethical, safe, and innovative AI. Is it possible to establish fair, ethical, and sustainable educational practices in the age of AI within the academic context? Under what regulations? The AI-ethics-HE triad, through these lenses, presents an encouraging yet still underexplored landscape for study and analysis.

This study aims to analyze recent research on normative and ethical frameworks governing the use of AI in HE institutions.

The literature search was conducted in the databases of Scielo, Redalyc, Scopus, Dialnet, and Web of Science. Additionally, AI tools such as Consensus were used to identify references. The search equations used to obtain significant results were: (“challenges” AND “promises of AI”); (“ethics” AND “AI” AND “higher education”); (“ethical considerations” AND “AI”); (“practicum” AND “AI”).

The main inclusion criteria for the documents were: articles published from 2020 to the end of May 2024; in Spanish, English, or Portuguese; Studies that include empirical research at the HE level. The exclusion criteria were articles without full-text access and/or that do not narrate experiences from the ethical field applied to education.

The initial search identified 256 records, of which 65 were selected. Among these, 14 were duplicates and 3 were not open access, leading to their removal. The remaining documents were processed using Mendeley and Rayyan for analysis and categorization. Figure 1 illustrates the methodological model based on the PRISMA statement.

Figure 1
Flow diagram of the PRISMA model

In the Rayyan platform, 48 records were exported for a second stage of screening and classification of the articles. Of the total, one was a duplicate, eleven were excluded, and eight passed to a second review; the rest were retained. Additionally, eight papers from other sources of information were included, while another eight were discarded for not being in the educational field. The labels used for the classification of the articles in the Rayyan platform were ethical use, didactic models, ethical challenges and risks, academic integrity, and regulatory frameworks.

Finally, 28 articles were selected, which were processed and categorized in Excel and then analyzed based on the categories: ethical challenges and risks, normative-regulatory frameworks, ethical training, and didactic models; these same categories were considered in the qualitative analysis software MAXQDA.

The synthesis of the main results is presented in Table 1 (see Annex), organized alphabetically. This table provides a summary of each study’s contextual information, including reference, methodology, study location, keywords and labels.

Figure 2 illustrates the evolution of scientific production on this topic between 2020 and May 2024. The year 2023 stands out, with a notable increase in research publications, totaling 17 works, coinciding with the impact of ChatGPT’s launch on November 30, 2022. In 2021 and 2024, four publications were recorded each year, while in 2022, only two works were documented, and in 2020, just one study was published.

Figure 2
Temporal distribution of articles

Scientific production is mainly concentrated in Spain and Ireland, which lead as the countries with the highest frequency of studies on the phenomenon, with a total of four publications each. They are followed by the Netherlands and Sweden, with two publications each. Other countries with one publication each include: Bangladesh, Canada, Chile, Croatia, China, Costa Rica, South Korea, Ecuador, Italy, Nigeria, Peru, the United Kingdom, and Switzerland. Additionally, 12 studies were identified where the geographical context was not specified

The analysis of the methodological approach provides insight into how research on this topic is currently being conducted. According to the data presented in the Annex, bibliographic analysis studies are the most prevalent, surpassing those with quantitative, qualitative, or mixed approaches. Specifically, 17 articles (58.6%) correspond to literature reviews, whether systematic or unsystematic. The remaining 42% is distributed among mixed studies (description of an experience based on a model), with each representing 3.4%.

Through gradual immersion in the material, using the content analysis technique (Ruiz Bueno, 2021), it has been possible to identify four central issues where attention is focused in ethical and normative terms: ethical challenges and risks, normative-regulatory frameworks, ethical training, and didactic models.

RQ1. What are the ethical challenges and risks that have been detected in universities? (references on this point are shown in Figure 3).

The use of AI in teaching, learning, and assessment processes presents challenges that have been recognized in recent studies worldwide, both from the perspective of teachers and students. Its impact on HE is considerable, generating both opportunities and risks (Chan & Hu, 2023). For example, cases are presented on how university students are adapting to the use of AI in their learning (Cisneros et al., 2023). Specifically, they warn that while AI facilitates access to information and resources, it can also result in an overwhelming amount of information when there is a lack of critical skills to assess it. Even with the prevalence of AI tools that are constantly growing, students sometimes engage in academic plagiarism (Comas Forgas et al., 2023). The ease of generating content leads to this tendency and raises questions about academic integrity, as well as the originality of student productions.

Figure 3
References associated with ethical challenges and risks

Without overlooking the opportunities offered by AI, such as the personalization of learning (Cordón García, 2023), potential risks for individuals and societies are identified. Indeed, there is talk of the "dehumanization" of education, coupled with a dependence on technology. Hence the importance of incorporating ethics into training that contemplates the use of AI in a fundamental way (Guissepe Hernández et al., 2021; Huang et al., 2021). We also find reports showing that sometimes anxiety and lack of self-efficacy in the use of AI by students (Kwak et al., 2022) can limit its adoption and effective application in specific environments. In these terms, an ethical awareness of AI is conceived, influencing the behavioral intentions of those who use it.

In summary, it is possible to emphasize that, although AI offers significant opportunities to improve teaching, learning, and assessment, it is essential to address ethical challenges, academic integrity, and the adequate preparation of students to mitigate the risks associated with its use. This also involves teachers and implies a proactive approach to training and regulating AI in educational contexts. Indeed, research reports on the topic (Ivanov, 2023; Vélez-Rivera et al., 2024) warn of the growing relevance that the AI-ethics binomial is gaining in universities, both at the general institutional level and in specific careers/areas.

RQ2. What are the normative and ethical measures that universities have applied or suggested regarding AI?

A significant number of documents review the need for normative frameworks and/or regulations on the use of AI in HE (Figure 4). It has been possible to recognize numerous initiatives around normative and ethical measures that have been proposed or applied by universities around the world in recent years. The European Parliament agrees that the implementation of an Academic Ethics Code provides a normative framework that establishes principles and values with the premise of guiding conduct in relation to the use of AI, both by students and teachers (Cornejo-Plaza & Cippitani, 2023). In this sense, universities have defined acts such as plagiarism and data falsification, among other dishonest practices, as violations of this code, with established consequences for such infractions.

Figure 4
References associated with regulatory and normative measures

Policies have been proposed to guide the ethical and responsible use of AI in academic environments (Aler et al., 2024). These policies address the need to ensure that AI technologies are used in a way that respects privacy, equity, and transparency. Therefore, it is expected that universities establish clear rules on the use of these tools, while providing guidelines for their integration into the educational process.

Hence, the relevance of advancing in AI training, with the implementation of workshops and courses that address the importance of academic integrity, plagiarism, citation, and good research practices (Cordón García, 2023). Aware of this, several Spanish institutions have developed mandatory modules in the curriculum, for first-year students to train them in digital ethics.

Regulatory proposals at the university level are not restricted to the teaching field, as the European Union has also established Research Ethics Committees (Piedra Alegría, 2023). The proposal is to review and approve research projects before execution to ensure ethics in data handling and the protection of participants. Yusuf et al. (2024) suggest the implementation of ethical impact assessments in projects involving AI, reviewing the social and ethical implications, which gives relevance to normative frameworks at this point.

In addition to the previous proposals, it is intended that there be regulations to generate inclusive environments based on AI, which encourage the participation of marginalized groups (Cox, 2021) through actions such as regulations for admission and scholarships to ensure equality and accessibility. Another aspect analyzed is university social responsibility and AI, so the principles of FATE (Fairness, Accountability, Transparency and Ethics) are examples of initiatives linked to academic and research activities for the well-being of the community (Henry & Oliver, 2022). All this achieved through extension and student and community collaboration projects, which encourage universities to be transparent in the use of AI algorithms and the resulting decision-making.

Data protection and privacy are another aspect to be guaranteed by institutions when incorporating AI, so regulations are aimed at ensuring that universities comply with these measures (García Peñalvo et al., 2024) through adequate management and safeguarding of information. In this regard, the evaluation of policies around the use of AI (Moya & Eaton, 2023) becomes relevant, so universities are invited to periodically review and update their regulations to support their effectiveness in the face of new challenges.

Along with the above, universities are encouraged to investigate the social, economic, environmental, and cultural repercussions of AI technologies to understand their impact on society (Chan, 2023). This synergy between research, teaching and community engagement can help institutions anticipate problems and adjust their policies based on findings.

RQ3. How is the ethical use of AI interpreted in educational processes? (Figure 5)

The findings obtained have allowed us to thoroughly understand the "ethical" implications of the use of AI in HE; this has triggered the need for a regulatory framework. Specifically, this question examines the current landscape of university policies regarding the use of AI. Of particular interest are Asian universities, which present a unique context for observing the institutional response to AI (Dai et al., 2024). It is expected that the findings will provide useful information on future directions for policy development in such universities as well as globally.

Figure 5
References associated with the ethical use of AI in educational processes

Academic integrity forms an essential pillar in education, in general, and in the training of professionals, in particular, especially in a world where AI plays an increasingly significant role. Multiple tools, including the well-known ChatGPT, are present in the training of occupational therapists (Avello-Sáez & Estrada-Palavecino, 2023). In this scenario, it is worth paying attention to academic integrity to avoid undue dependencies and promote authentic learning that will be required in professional performance; among them, the ability to interpret AI decisions in line with ethical criteria (Yu & Yu, 2023). This must be worked on from the beginning of university training, with strategies that reinforce academic ethics; as warned in Social Work and Social Education degrees (Comas Forgas et al., 2023), academic plagiarism can prevail among students.

Undoubtedly, there is an inescapable responsibility in the implementation of AI at all levels of the educational system, particularly in HE (Mora Naranjo et al., 2023), where an ethical foundation, along with academic integrity is essential for its genuine acceptance and sustainability.

RQ4. What are the didactic approaches or models that have been implemented in universities for the reliable use of AI? (Figure 6)

The growing integration of AI in HE has sparked debates around didactic models and the ethical use of these technologies. The educational models that are encouraged refer to an effective integration of AI into educational practices (Chan, 2023) through specific strategies for their use in various educational contexts. This is within the framework of academic integrity standards to be defined with comprehensive policies and guidelines that are as clear as possible. Benefits for learning are recognized, while concerns revolve around the balance of autonomy/dependence, as well as originality/plagiarism.

Figure 6
References associated with didactic models implemented in universities

One of the tools that has gained the most popularity in various fields, including HE and professional development of graduates, is ChatGPT. Given the importance of educating students and professionals in the proper use of these tools, guidelines are offered to promote the ethical use of this conversational AI to complement, and not replace, human competencies. This is in line with initiatives to foster a culture of academic integrity and ethical responsibility (Gallent Torres et al., 2023), particularly in the use of advanced technologies.

The concept of Education 4.0 (Gibert Delgado et al., 2023) has also been coined, referring to an approach that innovatively incorporates AI with the purpose of promoting the development of digital competencies and critical thinking in future professionals, in line with the challenges of an increasingly digitized world. Even ethics must constitute a core part of educational programs that incorporate AI, so that professionals are prepared to handle the ethical dilemmas that may arise in their work environments (Huang et al., 2021). In this context, it is expected that systems maintain high ethical standards (Aler et al., 2024), through transparent and fair evaluations, aware of the potential biases that may arise when including AI.

To summarize the relationships between the categories analyzed in this study, a co-occurrence map of codes is shown (Figure 7), which represents the ethical use of AI and its intrinsic relationship with solid normative requirements for the responsible, reliable, and transparent use of technology. The need for adequate regulations to mitigate risks is evident, through didactic models that promote a balance between technological innovation and the ethical principles that govern its use in HE.

Figure 7
Co-occurrence of codes in the documents

After conducting a descriptive analysis of the impact of the selected articles and with the aim of identifying possible trends in scientific production over the past five years (2020 - May 2024), a study of the relationships between their keywords was carried out. For this, the network analysis technique (Knoke & Yang, 2008) was applied, employing the VOSviewer program. From the 28 studies in the sample, 90 keywords were extracted, each with a co-occurrence index of at least two times. In the labeled relationship map (Figure 8), the nine clusters or keywords nodes are linked based on their degree of similarity, showing how they interrelate with each other.

The red cluster focuses mainly on research approaches that address the relationship between ethics and technology, especially in the fields of research involving human beings, such as education and medicine. Interest in using these technologies in educational environments is increasing, highlighting the role of digital platforms both in communicating findings and in integrating the educational community.

The green cluster focuses on the use of technology in HE to help promote inclusion at this level, especially for individuals with disabilities. It emphasizes the development of assistive technologies to improve inclusive education, facing ethical challenges in their design and accessibility to avoid inequalities.

The blue cluster highlights the challenges posed by digital transformation in the research and education sectors, underlining the role of AI as a driver of change. It emphasizes the need to ensure autonomy and academic integrity, adopting a human-centered approach.

The yellow cluster addresses the transformation that AI and big data analysis are causing in the educational sector, particularly in optimizing teaching and learning processes. These technologies allow for content personalization, adjusting to the specific needs of each student, with a particular emphasis on the use of social robots to enhance educational experience.

The purple cluster encompasses the challenges and opportunities that AI models offer in the educational sector, with a focus on academic integrity and learning assessment. The implementation of this technology has increased concerns related to plagiarism, requiring the creation of innovative assessment methods that discourage this behavior and encourage responsible use.

The turquoise cluster covers the legal field, emphasizing the role of AI in creating digitalization and automation tools that, in the educational field, offer multiple opportunities to improve accessibility and adapt teaching to individual needs. However, it also highlights the importance of assuming responsibility in the implementation of these technologies to reduce biases that could unfavorably condition learning outcomes.

The orange cluster analyzes the relationship between the use of algorithms, online supervision, and computer systems in HE laboratories, focusing on the difficulty of developing and using these technologies in an ethical manner.

The brown cluster refers to data and algorithm-based decision-making, also considering possible adverse effects, such as ethical implications and biases that may arise.

Finally, the pink cluster examines the impact of AI on access to and use of academic resources, while addressing the adaptation and privacy challenges that come with its implementation.

Figure 8
Labeled relationship map

Considering the size of the keyword labels and the color of the different approximation zones, the relevance of the studies is observed through the density visualization map in Figure 9.

Thus, in the central zone of the map (yellow color) are, due to their importance and co-occurrence, the keywords belonging to the research line related to AI, HE, and ethics, among which are: AI, high education, ethics, online proctoring, research article, ethics research, disability, among others.

On the other hand, the peripheral zone of the map (color close to green) shows the lower density of co-occurrence of the key terms. More specific expressions are located around which the analyzed research has been focused. Words such as: academic resources, automation, digitization, privacy, students, distance-based higher education, data set, algorithm are distinguished.

Figure 9
Density Map