1 Introduction

The world has changed since the release of OpenAI’s generative conversational tool at the end of 2022 [5]. The consequent concern about regulation and ethical behavior in the face of Artificial Intelligence (AI) has become the order of the day worldwide [7]. However, little attention seems to be paid to critical education for AI developers, especially in the Global South. UNESCO itself [9, 10] has published reports on the impacts of AI on education at basic levels and on society at large but has not addressed the professional education of those who create and employ the algorithms themselves. In addition, studies such as [4, 6, 14, 16] and [11] are at the frontier of the discussion on ethics, education and AI, but address issues outside the Brazilian context.

In the various areas of Computing, higher education is the natural path for both scientists and technical professionals who work in AI. In order to focus on education as a possible way of mitigating the risks and negative social impacts of AI, this article maps out the humanistic subjects offered in bachelor’s degree programs in the field of Computing at main Brazilian universities and classifies their approaches.

Despite the worldwide impact of the social challenges posed by recent generative technologies, one might consider that the discussion of ethics in the field of Computing has been going on for more than 20 years. Not only was the Association of Computing Machinery’s (ACM) code of ethics [1] published in the 1990s, but some academics have been dedicated to teaching this subject long before the contemporary phenomenon of AI [8], even presenting the same type of dilemmas, such as the reproduction of racism, the misappropriation of intellectual property, among others. However, through interviews conducted in this research, it was possible to gather evidence that these efforts have been and continue to be individual, rather than systemic or structural. Although there are groups concerned with the social impacts of AI within Computing, they are still small and their initiatives are insufficient in the face of current challenges.

There are more than 50 million places offered by around 22,000 bachelor’s courses in Computer Science, Information Systems, Software Engineering, and Computer Engineering registered with the Ministry of Education (MEC). This data demonstrates the magnitude of the area and reinforces the relevance of the discussion on the critical education of these students.

Considering the size of the universe to be researched, it would be beyond the scope of this work to analyze all the courses. the authors, therefore, opted for a subset of 17 universities, chosen for their importance in terms of publications in the area of AI, verified on the Scopus platform and in a Fapesp publication [12], and a ranking system, presented by the specialized publication Folha de São Paulo University Ranking 2023, as detailed in the Sect. 3 section.

In addition to this introduction, this article is composed of five other sections. The second section presents aspects of the field of Computing related to ethics. The third section presents the method used to collect the data. The next section discusses the quantitative and qualitative results of the subjects. The fifth section systematizes the syllabuses according to their classifications. Finally, the final section presents some of the authors’ conclusions.

2 The Construction of the Idea of Ethics in Computing

Professional activities have always involved ethical issues. This goes back to the most ancient civilizations whose history is documented, as well as the rules of native peoples whose traditions are lost in time. Of today’s professions, the medical field has the oldest tradition of ethical values. This can be traced back to the Hippocratic Oath, for example, which deals with commitments such as secrecy and beneficence in the doctor-patient relationship. Created more than 2.5 millennia ago and introduced into Brazilian medical practice in 1771, the oath is still taken at the end of undergraduate medical courses. Perhaps that’s why the biological sciences have such well-structured practices and regulations, as well as a network of 890 Research Ethics Councils (CEP) throughout Brazil.

In the case of Computing, although some discussions about the ethical use of technology goes way back, systematic discussions in congresses and seminars date to end of the 1980s and begin of the 1990s [15]. In addition to the fact that there are no Research Ethics Councils in Computing, it is not customary for professionals in the field to submit their projects to other councils, which should be the case, especially when involving human beings. This means that there is no formal impediment to computer professionals carrying out their work in an unethical manner.

3 Data Collection and Analysis

Having explained the context of Computer Science education and its dimension in Brazil, this section details how the data was collected and analyzed in this research.

In order to build the sample, the universities analyzed were chosen from three sources: i) the article “Scientific publications on artificial intelligence” by Fapesp [12], ii) Folha de São Paulo University Ranking 2023 (RUF) and iii) searches carried out by the authors on the Scopus platform. These sources were chosen to combine the relevance of the academic output of these universities with their other characteristics, such as teaching, innovation, and internationalization.

After defining the sources, the ten institutions with the greatest relevance in each of them were selected. Once the data had been cross-referenced and overlaps excluded, the final sample was consolidated in Table 1, presented below. A detailed explanation of the selection process is described in the following subsections.

Table 1. Sample Composition. Source: Author’s elaboration, 2024.
Full size table

3.1 Fapesp and Scopus Rankings

According to the article “Scientific publications on artificial intelligence” [12], the 10 leading institutions in AI publications between 2014 and 2018 were, respectively: USP, Unicamp, UFPE, UFMG, UFRN, UFRGS, UFPR, UFSCar, Unesp, and UTFPR.

Although international rankings like Edurank [17] and Times Higher Education [18] evaluate more metrics to achieve universities classification, they provide a generic and global ranking on all areas. As this article focuses on AI, the authors simplified the criteria aiming at this specific area in Brazil.

Fig. 1.
figure 1

AI publications by author affiliation. Source: Plataform Scopus, 2024.

Full size image

Based on Scopus searches using the keywords “artificial intelligence”, “machine learning”, “deep learning” and “natural language processing”, the 10 Brazilian institutions with the largest number of publications on each topic were listed. After eliminating overlaps, 11 universities were added. They were: USP, Unicamp, UFPE, UFMG, UFRGS, UFRJ, UFSC, UnB, PUC-RJ, UFF, and PUC-RS.

It is worth noting that just by searching for “artificial intelligence” as a keyword, there were more than 65,000 articles published in Brazil between 1974 and 2024, i.e. in the last 50 years. As shown in Fig. 1, the universities with the most publications in the area are, respectively: USP, Unicamp, UFRJ, UFMG, UFRGS, UFPE, PUC-RJ, UFSC, UFF, and UnB.

3.2 RUF

According to the Folha de São Paulo University Ranking 2023 (RUF) [13], an annual assessment of higher education in Brazil that considers teaching, research, market orientation, innovation and internationalization indexes, the best-placed universities are, respectively: USP, Unicamp, UFRGS, UFRJ, UFMG, Unesp, UFSC, UFPR, UnB and UFPE.

3.3 Consolidating the Field and Building the Database

Once the sample of universities to be analyzed had been selected, the bachelor’s degree courses were chosen. During the initial phase of the research, the authors held conversations with researchers from USP’s Center for Artificial Intelligence (C4AI) about the AI field. A university degree in Computing was presented as the main qualification among AI developers in the private sector. In addition, Resolution No. 5 of November 16, 2016 [3], which establishes the National Curriculum Guidelines for undergraduate courses in the area of Computing, states that the area is composed of the following courses: Computer Science, Information Systems, Computer Engineering and Software Engineering. These were the bachelor’s degree courses chosen by the authors for analysis.

In order to identify and analyze the subjects related to ethics and social sciences offered by these 17 universities in the four bachelor’s degree courses described above, the authors created a database containing the following variables for each subject: university where the course was offered, institute or faculty, course, name of the subject, whether it is obligatory or optional course, and syllabus.

To search for the subjects, the authors used the Curricular Matrices available on the official websites of the faculties and institutes or the Pedagogical Political Projects, depending on which document was more up-to-date. The subjects were searched for using the following keywords: ethics, society, social, Sociology, diversity, human rights, humanism and humanities. The terms were chosen from the bibliography, Resolution No. 5 of November 16, 2016, and a pre-analysis of the curricula.

3.4 Categorization of the Syllabuses

After creating the categories for analysis, adapted from the literature review by [2], each syllabus was read and classified by at least two authors independently. Both classifications were then checked, and, in cases of disagreement, the authors debated until a consensus was reached.

3.5 Semi-structured Interviews

To complement the understanding of ethics teaching in the context of higher education, three semi-structured interviews were carried out, one with a professor of ethics in Computer Science, one with a professor of ethics in Engineering, and one with a professor of ethics and head of a Computer Science Department. In all cases, the interviewees were part of the teaching staff of the universities that make up the sample studied.

4 Analysis of Ethics-Related Subjects in Computing Courses

This section presents two analyses. Firstly, considerations on the survey of ethics-related subjects offered in the courses evaluated. These are quantitative assessments. In the next subsection, qualitative analyses are made. The authors define “ethics-related subjects” as the set of subjects identified using the selected keywords: ethics, society, social, Sociology, diversity, human rights, humanism, and humanities.

4.1 Descriptive Analysis of Ethics-Related Disciplines

This section deals with the research analysis into the quantitative evaluation of ethics-related disciplines in the Computer Science, Computer Engineering, Software Engineering, and Information Systems courses of the 17 universities studied. To the assessment, quantity and the compulsory nature of those disciplines were considered. This analysis made it possible to understand the interviewees’ accounts of the ideological disputes in the field, which prioritize certain areas of knowledge to the detriment of others, and prompt a broader debate about the need for critical education of future AI professionals.

Although the research sample included 17 universities, not all have the four elected bachelor’s degree courses. Table 2 below shows the relationship between the courses and the universities where they are offered. It is important to take this information into account when analyzing the number of ethics-related subjects below, since the number of disciplines was calculated in absolute numbers and not proportional to the number of courses.

Table 2. Courses offered by each University. Source: Author’s elaboration, 2024.
Full size table

Based on a survey of ethics subjects and related topics (such as social sciences and human rights) in curricular matrices and Political Pedagogical Plans, a total of 204 subjects were found. Of these, 83 were part of the bachelor’s degree courses in Computer Science, 67 belonged to the Computer Engineering courses, 12 to the Software Engineering courses, and 42 to the Information Systems courses.

As seen in Fig. 2, UTFPR is the university with the largest number of ethics-related disciplines. It offers all four degrees, having 55 disciplines, with Computer Engineering having the most (19) and Software Engineering the least (4). At the other end, Unicamp and Unesp have only one discipline each, and none was found at UFPR.

Fig. 2.
figure 2

Number of disciplines related to ethics, per course and university. Source: Author’s elaboration, 2024.

Full size image

The disciplines were classified as compulsory or optional/ elective/ complementary by analyzing the above-mentioned documents. The optional/ elective/ complementary disciplines were identified in the Curricular Matrices and Political Pedagogical Plans and were considered as a single block here treated as “optional disciplines”. Furthermore, as non-compulsory subjects can be elective (compulsorily offered by the university, regardless of the number of applicants) or optional (they may not be offered), it was not possible to determine their classification in the surveyed data and whether they have been, in fact, offered or not. In the interviews, the authors found out that there are registered disciplines in university systems that are not, and have never been, offered.

As shown in Fig. 3, 132 disciplines are optional, which represents a large majority compared to the 62 compulsory ones. Thus, from the total of 204 ethics-related disciplines, only around 30.4% are obligatory, and it is up to the students individually to choose from the remaining 69.6%.

Fig. 3.
figure 3

Number of obligatory vs. optional disciplines per university. Source: Author’s elaboration, 2024.

Full size image

4.2 Analysis of the Interviews

The findings about the disparity between compulsory and optional subjects are aligned with the information from the interviews conducted with Computer Science and Engineering professors. According to what was reported, disciplines related to ethics, social responsibility, and social issues linked to technology are usually the result of personal initiatives by teachers who are interested in the subject and who, often in a self-taught way, begin to study philosophy in particular, incorporating these themes into their disciplines. They usually do this informally in technical classes, discussing real cases involving ethical issues, for example. Less often, they opt to formalize specific subjects on the topic.

When asked about the legal Resolution No. 5 of November 16, 2016, that makes humanistic teaching compulsory, the teachers and the head of department interviewed said they had no prior knowledge of it. In addition, the head of department said that, during a visit to inspect the course, the MEC did not analyze compliance with the aforementioned Resolution. In addition, the interviewees said that there was a certain amount of demise or even resistance from colleagues to the introduction of this kind of disciplines. According to them, there is an ideological dispute in the field about the essential nature of higher education for professionals in the field of Computing. Although not explicit, the neglect of ethics-related in favor of technical disciplines is based on a traditional and market-oriented discourse.

As for student interest in ethical issues, the interviewees described the profile of Engineering and Computing students as more introspective and less participative in class. For one of the interviewees, this attitude is due to the context of what is valued by companies as well as a lack of interest in social impact. However, the other interviewees attributed this behavior to typical characteristics of the social group and a more traditional hierarchy between teachers and students that is ingrained in these bachelor’s degrees, and not necessarily due to the discipline itself. Furthermore, one of the interviewees considered quota students to be more politicized and interested in discussions about ethics.

5 Analysis of the Syllabuses

5.1 Creation of Categories

Ethics teaching covers a wide range of concepts and has roots in various knowledge fields. Reading the syllabuses of the collected subjects showed themes related to the fields of Sociology, Philosophy, and Law. The categories used to analyze the syllabuses were adapted from the review by [2]. In this review, the articles were identified according to the presence or absence of an explicit description of the conception of ethics, and how these concepts appeared (exclusively or mixed) in them.

[2] present three main conceptions of ethics found in the literature review: Relative to Philosophical Frameworks, Computing Applications, and Social Impacts of Computing. In general, the authors state that these conceptions used to mention “normative ethical theories or situate ethics in relation to concepts of Computing, social impact and professionalism” [2] (p. 12).

About the philosophical conception, [2] ethics is identified as a Western philosophical theory, including issues such as consequentialism, virtue, and duty, as well as a discussion of decision-making processes. On the other hand, the articles that dealt with ethics from a computational perspective did so from debates close to the legal field, addressing issues of security, privacy, and responsibility, for example. The third approach focused on the relationship between technology design and its social impacts, as well as issues such as accessibility.

As mentioned in the Sect. 3 section, the selection of subjects from Computing courses was broader than just those specific to ethics. Sociological and humanistic subjects were also taken into account. For this methodological reason, an adapted version of the three conceptions of ethics by [2] was adopted. The Computing Applications and Social Impacts of Computing categories were partly merged and partly fragmented. On the one hand, the issue of social impacts was expanded to cover not only the relationship between Computing and society but also typical Sociology topics that go beyond technology, such as Social Science theories, racism, gender inequalities and others. On the other hand, all the topics related to the profession of future Computing professionals, including both legal and responsibility debates, were compiled as a second block.

Based on the inspiration provided by the work of [2] and the particular reading of the set of syllabuses in this research, three ways of classifying the issues linked to ethics and society found in Computing subjects were drawn up:

  1. a)

    Philosophical framework: Syllabuses that refer to theories, concepts, thoughts, and thinkers from the philosophical tradition; epistemology of ethics; human values; principles of justice, truth, good, freedom, etc.; mention of the notions of consequentialism (ethics based on consequences); deontology (ethics based on duties), and ethics based on virtues.

  2. b)

    Applied to Computing: Syllabuses that refer to ethics in relation to the concepts and areas of application of Computing; everyday dilemmas faced by professionals; responsibility of professionals towards the community and society; ethics in decision-making; best practices and procedures; quality and security of computer systems; quality of scientific knowledge production; topics related to privacy, intellectual property, legislation, and codes of professional ethics.

  3. c)

    Social issues: Syllabuses introduce the concept of ethics through the foundations of sociological thought and address the social impacts of Computing; reference to the social and historical character of the technological artifact; social repercussions and changes in how society is organized, such as in work, motivated by technology; the problem of reproducing inequalities, and stereotypes (of gender, race, class, sexuality, etc.); notions of risks and threats to the well-being of society; notions of citizenship.

5.2 Systematization of Syllabuses

All the syllabuses found in the Curricular Matrices and Political Pedagogical Projects of the four types of courses considered were analyzed based on the three categories mentioned in Subsect. 5.1. This showed 180 non-repeated disciplines. It is important to note that, of these 180 subjects, 48 did not have their syllabus available. Among those that did, there was a great disparity in the amount of information available. While some had detailed content in several descriptive paragraphs, others provided simplified generalizations in just one line. Thus, although there was one example with more than 200 words, the average was 50 words. With this in mind, about the classification, each discipline was placed in one to three categories, depending on the syllabus. Table 3 shows the following scenario:

Table 3. Courses offered by each University. Source: Author’s elaboration, 2024.
Full size table

As can be seen in Table 3, all three categories have a significant presence in all disciplines. However, disciplines that deal with social issues are more frequent and those with a philosophical framework make up a smaller group.

Furthermore, as shown in Table 4, the category “Social issues” appeared more often alone than the others. When combined, it is usually linked to the “Applied to Computing” category. Only 10 disciplines were categorized under “Philosophical framework” and “Applied to Computing”, and only eight had all three approaches.

Table 4. Classification. Source: Author’s elaboration, 2024.
Full size table

6 Conclusions

The emergence of Generative AI and the widespread use of AI-based technologies have intensified the debate on ethical issues, both nationally and internationally. The establishment of professional ethical values through education is an important step towards mitigating the negative social impacts generated by the use of AI tools.

The topic’s relevance is also demonstrated by the size of the field of Computing, which in Brazil offers thousands of courses and millions of bachelor’s degree places. For this reason, this article presents an introductory overview of ethics-related teaching in bachelor’s degree courses in Computer Science, Computer Engineering, Software Engineering, and Information Systems at the 17 main Brazilian universities.

Based on the bibliography, an analysis of university documents and semi-structured interviews, the authors found an important disparity in the provision of ethics-related subjects/disciplines between universities. If, on the one hand, the Federal Technological University of Paraná (UTFPR) and the Federal University of Rio de Janeiro (UFRJ) were the only institutions to have more than 50 humanistic subjects on their courses, 13 universities out of the 17 evaluated offered less than 10 subjects of this type; the State University of Campinas (Unicamp) and the São Paulo State University (Unesp) with just one and the case at the Federal University of Paraná (UFPR) with none.

A significant number of subjects had no syllabus, around 24%. However, among those that did, at least one reference to some philosophical framework, applications in Computing or social issues was identified. As in the review work by [2] this does not mean that clear conceptualizations of ethics were found in the syllabuses. It just shows that, in terms of the delimitation suggested in this work, it was possible to analyze the distribution of references to fields of knowledge and to some sets of themes that the disciplines refer to in order to address the problem of ethics and social dilemmas in the context of Computing education. The balanced distribution of categories suggests no predominance of one approach. One possible explanation is that, since there is still no “canon” for the teaching of ethics in the area, there is a plurality of approaches depending on the proposal of each subject. The semi-structured interviews, although only three, showed a tendency toward relative pessimism and frustration, with little belief in the possibility of greater academic space for such ethical discussions and devaluation of the subject in relation to technological aspects.

The authors recognize the limitations of the data, which does not allow them to draw any major conclusions about the state of ethics teaching in Brazil. This article was intended to record the first findings of a more extensive research agenda, which could expand both to a bibliographical analysis of the disciplines offered by universities and to an understanding of the way they are taught by lecturers and understood by students. In addition, it is proposed to deepen the gathering and interpretation of discourses on the importance of critical education in Computing and its relationship with the development of AI for the social good. Finally, encouraging scientific debate on the ethical education of future AI developers could stimulate the application of the Resolution No. 5 of November 16, 2016 and suggest approaches to make this education effective.