Sustainable Social Development through the Use of Artificial Intelligence and Data Science in Education during the COVID Emergency: A Systematic Review Using PRISMA

Abstract

In this paper, we aimed to investigate how sustainable development can be involved in educational contexts that use new trends in technology such as Artificial Intelligence (AI) and Data Science (DS). To achieve this goal, we conducted a documentary Systematic Review using PRISMA research to find and analyze applications of sustainable development in these aforementioned contexts. In the results, we explain how some applications of both AI and DS, including Big Data and Learning Analytics among others, offer alternatives to substantially improve the educational process by allowing either education to be personalized, learning to be predicted, or even possible school dropouts to be predicted. We found that the COVID emergency sped up the introduction of these technologies in educational environments. Nevertheless, the inclusion of new technologies to develop better processes in education is still in progress and will continue to grow. In conclusion, we identified and analyzed some of the main applications found in the literature regarding new computing technologies. AI and DS have introduced new learning and teaching methods to solve different context problems that promote sustainable development while making educational environments more dynamic. On the other hand, we observed that the divide in many countries will continue due to their economic and technological situations. We describe some of the challenges that the incorporation of these technologies will bring in the near future.

1. Introduction

Due to the progressive transformation of Higher Education Institutions (HEIs) and social dynamics [1,2,3], a new concept of society has emerged: the knowledge society [4,5]. This new concept integrates the importance of solving and considering new challenges regarding sustainable social development for which we must provide efficient and effective responses. The CIFE (Ciencia e Innovación para la Formación y el Emprendimiento/Science and Innovation for Training and Entrepreneurship) Institute in Mexico proposes the concept of a knowledge society in accordance with the challenges of Latin America. It is defined as a group of people who work collaboratively to solve context problems [6] via complex thinking and the coconstruction of knowledge by taking advantage of various resources such as information and communication technologies (ICT) [7,8,9,10]. In this new approach, the priority is not knowledge itself or technology but people and their development [11]. Facing the rapid evolution of this new society and the challenges derived from the COVID-19 pandemic as well as other types of viruses emerging, there are questions that should be answered: What are the challenges of Higher Education (HE)? How should HEIs operate in the present and in the future [12]? Innovation is an evident need in academic–cultural, pedagogical–didactic, technological–educational, transdisciplinary education [13,14,15,16], organizational, investigative, institutional welfare, permanent education, and social projection aspects.

The pandemic has expedited the need to turn to digital tools in education and other organizations in response to the situation faced by all nations around the globe [17,18,19]. This global situation has brought a need to enforce the use of technology to communicate and transfer information that has never been experienced before [20,21]. None of the organizations, including educational institutions, were prepared to change from presential to virtual, not only in terms of pedagogical aspects but also technical and technological matters. Individuals following a socioformation approach have proposed a concept for “talent” that fits into these new educational conditions, defining it as the set of actions that human beings must perform in order to improve their living conditions through the resolution of context problems. Context problems include the pandemic, among other global challenges. The socioformation approach is based on universal values and complex thinking such as critical, systemic, metacognitive, and creative analysis [22]. This concept can be used to overcome the new educational and society challenges. Even when this concept was issued before the pandemic emergency, it emphasized the importance of training people that can criticize, analyze, and argue to solve problems and who can work with others to overcome difficulties and contribute to sustainable social development [23,24]. Referring to quality of life, this approach considers the actions that are necessary to preserve life by taking into account individual social responsibility and also the collective responsibility to work for the continuous development of new talented people who can achieve organizational goals as well as to obtain results regarding the improvement in quality of life [25,26]. Individuals following a socioformation approach look forward to integrating knowledge in a multidisciplinary and transdisciplinary way since the approach promotes the integral development of people by executing socioformative projects with an ethical vision. They also look forward to being critical and reflective. On the other hand, this approach promotes the use and proper management of IT and other new technologies as partners. It is considered to be different from other approaches since it emphasizes the use of knowledge in educational contexts to solve real and significant context problems such as global warming; organized crime; human destruction due to drug use; poor nutrition in the population, mainly in children who suffer from anemia or are overweight; human violence in all social contexts; a lack of humanism in organizations; the emergence of new diseases such as COVID 19; etc. It also suggests that appropriate strategies should be designed and applied according to the requirements and that individuals should work in multidisciplinary environments to achieve optimal results [27,28].

In educational contexts such as universities, it is difficult to perform the actions that lead to environmental sustainability [29,30] mainly because of the lack of understanding of how to perform actions to achieve optimal results in behavior changing. One of the main reasons is the resistance to change, both in the academic structure and in management [31,32,33]. Nevertheless, there have been approaches to introduce innovative ways to transfer knowledge and the understanding of the concepts of sustainable development [34] through business simulation games [35].

Environmental sustainability and sustainable development are concepts that have been misunderstood because they have been treated as synonyms [36,37,38]. Therefore, efforts towards making the necessary actions to achieve either one or the other have not been clear enough in the literature [39,40]. It is pertinent to bring society together to spread knowledge and achieve social commitment and responsibility regarding the importance of caring for the environment but also caring for social development, which requires the efforts of all individuals. It is relevant to conduct research with the help of science [41] and other new technologies that allow for processes that facilitate the appropriate treatment of problems to be improved.

The use of Data Science (DS), for instance, is useful to successfully synthesize heterogeneous data from diverse sources to support complex analyses and the extraction of new knowledge [42,43].

To avoid confusing sustainable social development with the concept of environmental sustainability, it is important to point out that the socioformative approach includes sustainable social development as a fundamental element and that this, in turn, includes environmental sustainability as one of its elements. Therefore, sustainable social development is conceived as the actions that together lead to the achievement of a better life in terms of physical and mental health as well as good interpersonal relationships, access to medical services, education, and housing among others. Socioformation emphasizes the concept of sustainable social development and proposes collaborative work, inclusion, and metacognition as part of the transformation, generation, and cocreation of knowledge and the development of activities that lead to environmental sustainability [7].

On the other hand, regarding the current challenges, Artificial Intelligence (AI) is an efficient and effective tool that can be used to solve specific problems in society. The Royal Spanish Academy defines AI as a “scientific discipline that deals with the creation of computer programs that perform operations that can be compared to those performed by the human mind, such as learning or logical reasoning” [44]. In other words, it involves the study of how to make computers perform things that, at the moment, people are better at performing [45]. It is considered to be a discipline that arose before Data Science. Both have contributed to the educational field, and there are extensive examples of their use: the ubiquity of digital devices in schools, online learning, the availability of digital material in different media, and the ease of communication in what is practically real time, just to mention a few [46]. DS is a discipline that both public and private institutions are currently involved in [47,48]. It is composed of a set of principles, problem definitions, algorithms, and processes for the extraction of useful patterns from large data sets [49]. Computer science and statistics disciplines such as data analysis, machine learning, data mining, and pattern recognition are included in DS [50]; however, different application domains are emerging, and education in sciences is one of them [51]. Environmental sciences is another one too [52]. Some subdisciplines such as data mining and machine learning stand out from DS; the first one is regularly used to perform structured data analyses for commercial use, and the second is used to design and evaluate algorithms for the extraction of data patterns [49].

The use of a vast variety of disciplines in all educational areas to face the new challenges brought by the pandemic and knowledge society helps to fulfill the gaps in education and allows for the development of research or new educative innovations as proposed by sustainable social development. The integration of multiple areas allows us to achieve high levels of quality through collaboration. Some of the most relevant and modern technologies were analyzed in this research and will be presented in the results.

Finally, this paper addresses the following research questions:

Q1. In terms of conducting research, what regions related to the use of AI and DS applications in education stand out as alternatives to develop better and more available ways to transfer knowledge as well as consciousness for sustainability?

Q2. What are the main educational tendencies when using DS and/or AI in an educational context?

Q3. What are the main uses of AI and DS for educational purposes?

2. Materials and Methods

2.1. Type of Study

We carried out a documentary analysis by using the systematic approach methodology Preferred Reporting of Items for Systematic Reviews and Meta-Analysis (PRISMA) [53] (Table S1 PRISMA Checklist, see Supplementary Materials) to determine how AI and DS have been used in educational contexts to provide the most effective ways to transfer big amounts of information and to make more efficient educational scenarios. This type of research consists of selecting and analyzing the information and knowledge existing on a subject to determine relationships, categories, differences, and stages. We determined what type of research papers to select, where to find them, what criteria to use when searching, and what document analysis was necessary to reach our established goals.

2.2. Computing Areas Analyzed

To determine the areas where these technologies are mainly used, we conducted an exhaustive analysis of the most representative literature related to AI and DS applied in education, and even more specifically of those that referred to sustainable social development or related topics. It was complicated to categorize which applications belonged to AI or DS since both disciplines overlap and also because one can depend on the other depending on the approach; i.e., AI can be used without DS, but DS will depend on AI if it uses automatic learning algorithms or deep learning. In

Table 1. Categories.

Computer Discipline	Main Use
Artificial Intelligence	Adaptive learning
	Intelligent Tutoring Systems
	Tutorial Dialog Systems
	Q&A Systems
Data Science	Recommendation Systems
	Learning Analytics
	Personalized learning
	Predictions
	Behavioral analytics

, we show the categories of the types of uses of these technologies found in the literature.

2.3. Research Papers Selection Criteria

Once we detected the kind of computer discipline and its use in educational contexts, we proceeded to do the following:

• We searched for original published books and research papers within databases such as Web of Science (WoS), Scopus, Science Direct, Scielo, Redalyc, Latindex, IEEE, and Google Scholar. We selected papers published in indexed journals and books of recognized publishers, research centers, or prestigious universities.
• We searched for research papers in the databases using several keywords and combinations of them, both in English and Spanish. The search process was conducted manually. These words and combinations were: “AI”, “Artificial Intelligence”, “DS”, “Data Science”, “education”, “analytics”, “learning analytics”, “desertion”, “prediction”, “sustainability”, “education”, “adaptive learning”, “social sustainability” “AI and COVID”, “Artificial Intelligence and COVID”, “DS and COVID”, “Data Science and COVID”, “education and COVID”. The articles were first selected because they included AI or DS technologies applied to education. Then, it was important that AI or DS were relevant to the educational process. Empirical studies were preferred but not exclusive for selection.
• We mainly selected research papers that were published between the years 2014 and 2021 that were related to the use of the computer disciplines as explained above. The time frame that we selected was the five years prior to the first identified cases of COVID. This was due to the fact that we expected to find a relative increase in the importance of the use of AI and DS in educational processes as the pandemic period pushed all sectors of the population to use alternative ways to communicate that depended on new technologies.

Figure 1 shows the literature selection process.

2.4. Limitations

This article is a documentary analysis that used the systematic approach methodology in order to establish how AI and DS can contribute to knowledge in various educational environments. In this sense, the literature was reviewed in a manual way by using Excel spreadsheets to organize the article information. This research involved studies published five years prior to the appearance of COVID and during the pandemic until 2021 to visualize the progress of these technologies and how they helped education. The search was limited to 8 international databases. In addition, the search was conducted in just two languages, English and Spanish, whereby most articles related to the studied topic were in English. Further research could consider some additional languages to include other possible research not written in English or Spanish.

3. Results

3.1. Statistics of Research Papers Selected for Analysis

Table 2. Research papers summary.

Research Papers	Total	About the Subject	Contextualization orComplementary
Theoretical papers	82.76%	87.50%	12.50%
Books	13.79%	62.50%	37.50%
Others	3.45%	50.00%	50.00%

shows the total percentage of the analyzed and classified research papers and their context.

Table 3. Summary of research papers classified by continent.

America	Asia	Europe	Oceania
35.76%	17.24%	40.83%	5.17%

shows the distribution of the research papers by continent, whereby the most significant studies concerning the proposed research topic were conducted in Europe. In total, 81.03% of the total selected research papers were written in English, and 18.97% were written in Spanish.

Table 4. Summary of research papers classified by year of publication.

2014	2015	2016	2017	2018	2019	2020	2021	2022
3.80%	4.29%	4.93%	6.07%	8.68%	13.00%	21.02%	32.03%	6.15%

presents a percentage summary of the research papers selected by year from 2014 to 2022.

3.2. Artificial Intelligence and Its Use for Educational Purposes

A particular strength of human intelligence is its capability to adapt to diverse conditions such as environmental conditions and adjust or change one’s behavior through learning experiences [46]. AI has been developed from its creation, up to date, via four system approaches: (1) thinking humanly, (2) acting humanly, (3) thinking rationally, and (4) acting rationally, which has resulted in a confrontation between those that are centered on human behavior and those centered on rationality. The first two are empirical and involve hypotheses and confirmations with experiments, while the rest involve the combination of mathematics and engineering [54].

One of the areas of AI in education is adaptive learning; according to ITESM [55], this kind of learning involves the customization of the content according to the strengths and weaknesses of each student, which offers different possibilities. Some of the most relevant adaptive learning and evaluation systems are: (1) adaptive tests, which are based on computerized adaptive tests (CAT) and are built with algorithms that produce optimal tests for each student; (2) adaptive tutorials that allow students to interact with the material by simulating a task or objective; and (3) cognitive tutors that use AI algorithms to simulate the behavior that a human tutor should have.

Massive and Open Online Courses (MOOCs) are now widely used. Baneres, Caballe, and Clariso [56] presented an analysis of three of the most relevant MOOC platforms: Coursera, edX, and Canvas Network; with their results, they demonstrated that the current analytics implemented by these platforms are not prepared to support Intelligent Tutoring Systems (ITS). They used six evaluation categories to conduct their analysis: qualitative, quantitative, student comparison, learning resources, student achievement, and retention. In the article, they also presented an eLearning development platform called ICT-FLAG, with the purpose of providing ITS and eLearning tools with innovative services to benefit students, instructors, administrators, and academic coordinators through formative assessment tools, Learning Analytics, and gamification.

Within the same ITS, there are the Tutorial Dialog Systems. Ezen-Can and Boyer [57] compared two of these systems, one based on a supervised dialogue act classifier and another that depends on an unsupervised classifier. This study was conducted with 51 computer science students and demonstrated that: (1) both versions of the system achieved similar gains in learning, as well as in user satisfaction, and (2) some characteristics of the new students were highly correlated with the perceptions of their experience during tutoring. They concluded that because the students’ motivation was strongly correlated with the results, adaptive systems that adjust their strategies according to the motivation of the student are a promising proposal to improve the systems of tutorial dialogue.

Atapattu, Falkner, and Falkner [58] mentioned that question and answer (QA) systems have been explored within educational contexts to facilitate learning and that most of them have focused on text-based responses. Therefore, they proposed to present them as a conceptual map in order to foster meaningful learning and knowledge organization. They conducted a random experiment with 59 university students in computer science and obtained statistically significant results for learning gain when concept maps of each question were provided to the students. It is important to note that they used Natural Language Processing (NLP) tools developed by Stanford University as well as free software CMap Tools.

Another research topic in the literature is peer learning. It has been a valuable complement to formal education. Potts, Khosravi, and Reidsema [59] said that if peer learning is performed without guidance or context, it may not be as useful as it could be. To minimize this risk, they used an open-source Recommendation in Personalized Peer Learning Environments (RiPPLE) platform, which makes recommendations for peer study sessions according to the availability, competencies, and compatibility of the students. In their work they provided content from a repository of multiple-choice questions to facilitate and motivate thematic debate and assist in the process, all with the help of a knowledge-tracing algorithm and a Gaussian scoring model that made it possible to select questions that promoted relevant learning and met the students’ expectations.

3.3. Data Science and Its Use for Educational Purposes

A large amount of data are generated every day. They can be digitally stored thanks to technological advances, which have supported the proliferation of massive storage systems. Such data can be structured or unstructured. Most of the data are generated with many technological communication tools such as social networks, forums, and blogs, among others. This is one reason why all types of organizations need to seek new technical solutions to manage this vast amount of information. For example, different users may perceive that other organizations offer a better and personalized service, which will improve their relationship with these organizations.

Big Data can produce enormous benefits for society, such as advances in medicine, education, health, transport, and urban development [60] without using users’ personal information being disclosed in many cases [61]. The emergence of Big Data provides excellent opportunities to change educational systems and programs. In addition to other technological advances, they lead to innovation in education [62]. One of the main characteristics of Big Data is that the information is continuously generated and seeks to be comprehensive, far reaching, flexible, and scalable. It allows for the implementation of new approaches to the generation and analysis of data that allows for questions to be constructed and answered in different ways [63].

Within the educational context, according to Ong [64], it is important to consider the response time of the real-time data generated by students such as data on learning activities in online courses, attendance, and the various statistics generated during the teaching–learning process. In general terms, organizations have an interest in making decisions in real time as it gives them a competitive advantage. Big Data analytics refers to the use of data, statistical analyses, and the creation of explanatory and predictive models to visualize the data and obtain information that will help in decision making.

On the other hand, new computer and software development technologies have enabled the creation of tools to be used in the teaching–learning process, including virtual learning environments (EVA) and learning management systems (LMS). These tools leave digital fingerprints of the users, such as the number of times they access a specific platform, the sections they access, and the length of time they stay on it, among others [65]. This digital fingerprint is generally stored in log files which, depending on the number of users, can be large and can later be analyzed with computational algorithms according to what the user wants to know, interpret, or predict.

According to Amo and Santiago [66], Learning Analytics consist of a set of statistical techniques applied to the qualitative and quantitative data collected and generated during the teaching–learning process in different educational contexts. The purpose of this information is to be interpreted through quantitative approximations to understand, explain, and predict student behavior with the intention of improving that context. According to the authors, the purpose of Learning Analytics can be encompassed and interrelated in four levels with respect to their phases and execution times: (1) Explanation, which corresponds to the visualization of the data of the present and the past. (2) Diagnosis, which involves the analysis of the visualizations and is carried out in the same way with new input data instead of the old data. It requires analytical literacy to develop the extraction, conclusion, and diagnosis of the visualizations. (3) Prediction, which involves the interpretation of the analysis to make a prediction based on the collected data. And (4) Prescription, which involves the interpretation of the present and future predictions or analyses. Learning Analytics define the actions that will take place once the visualizations and results have been analyzed. The role of the teacher in this new era is of utmost importance as they need to develop analytical skills to comply with the phases mentioned above to be able to customize or adapt both virtual learning environments and learning itself. The collection of data on student interactions, through the different quantitative resources available, can generate the technological tools that precisely help this personalization of learning.

Cen, Ruta, and Ng [67] mentioned that the flexibility and scalability that new computational technologies allow in terms of the processing, data analysis, and extraction of new knowledge and meaningful information from educational data can benefit students, teachers, and the entire educational ecosystem in general. The authors found four uses of Big Data that can have a significant impact on education: (1) the prediction of school performance; (2) a recommendation system based on such predictions; (3) data-based Learning Analytics; and (4) personalized learning. They presented a case study in which they demonstrated the effectiveness of different interaction patterns found in time series in a collaborative learning environment.

Govindarajan, Kumar, Boulanger, and Kinshuk [68] found empirical evidence that their approach to Learning Analytics reduces the failure rate of courses and improves the success rate of students. In their work with simulated student data, they adapted and combined the characteristics of three algorithms to improve the learning process: swarm intelligence; Naive Bayes; and collaborative filtering for clustering, prediction, and recommendations. It is interesting to analyze what they propose as future work since they intend to expand their study to integrate and explore other prediction and recommendation algorithms, predict and reduce student dropout and retention rates, and conduct experiments in real time by using their proposed framework.

In the work of Yang and Huang [69], they presented the research conducted and experience acquired by applying Big Data analytics to the educational cloud (EduCloud). They focused on the infrastructure to carry out the activities of data collection, cleaning, storage, query, analysis, and visualization to make the proposed platform suitable for teachers and students in Taiwan. This initiative has three dimensions: digital infrastructure, open resource services in the cloud, and a learning model with innovative ideas. In their future work, they will analyze the administrative processes of the schools, the teaching–learning process, personalized learning, the adaptive assessment of remedial classes, and student behavior in MOOCs.

Collaborative learning can be subject to analysis with Big Data techniques, as was the case with the work of Lu et al. [70], in which this analysis was used to measure the rate of student participation in a collaborative learning environment. For this purpose, they developed a collaborative programming tool called the Software Project Development and Integrated Learning Environment (SPDI Learning Environment), which allowed computer science students at the National Central University and the University of Technology, both in Taiwan, to develop their coursework collaboratively. To perform this analysis, they implemented a point-based engagement measurement algorithm in parallel with the MapReduce framework as well as a user interface for students in the web-based development environment, and another for instructors to use the visualization panel. It is worth mentioning that the Big Data design principles used were (1) data analytics and behavioral analytics, (2) hidden patterns and unknown correlations, and (3) market trends. The authors did not present their results; they showed the different phases of their project and indicated that only the events generated by the mouse and keyboard produced the data. With this, they were able to investigate the behavior of the students by analyzing the flow data of these events. They also affirmed that the application of Big Data technology in the analysis of learning can improve the curricular model, adjust responsibility in real time, and identify who has difficulties in what steps during the courses.

Yu and Wu [71] described some typical examples of the use of Big Data in education by focusing on the mining of educational data with the aim of analyzing the data generated to solve various problems of educational research such as predicting the future learning of students, discovering or improving the models of the content domain, studying the effects of the pedagogical support of learning software, and advancing scientific knowledge on student learning. Among the types of usages of Big Data shown were (1) prediction of academic performance, (2) presentation of performance, and (3) understanding students’ learning activities. They presented some examples as a description of usages without results and concluded, among other things, that personalized learning is a trend for the future of education since the needs of each student can be satisfied through a recommendation system that personalizes learning resources by providing an adequate route. They also mentioned that administrators can better organize learning resources, direct educational reforms, and take appropriate measures.

In an article by Zhang et al. [72], the authors presented the concepts and design of a platform through an application scenario that used Big Data technologies for online education, with the intention of improving educational quality through the analyses generated during the use of that platform; they did not present results since they were only describing the characteristics of the platform and the expected benefits of its use.

3.4. Summary of the Most Representative Literature Related to AI and DS Applied to Education

Table 5. Main use of AI and DS in education.

ID	Main Use	Title	Date	Author
AI	Adaptive learning	An Introduction to Artificial Intelligence	2017	Ertel, W. [46]
AI	Q&A Systems	Artificial Intelligence: A Modern Approach	2016	Russell, S.; Norving, P. [54]
AI	Adaptive learning	Reporte Edu Trends	2014	ITESM [55]
AI	Intelligent Tutoring Systems	Towards a Learning Analytics Support for Intelligent Tutoring Systems on MOOC Platforms	2016	Baneres, D.; Caballe, S.; Clariso, R. [56]
AI	Tutorial Dialog Systems	Tutorial Dialogue System for Real-Time Evaluation of Unsupervised Dialogue Act Classifiers: Exploring System Outcomes.	2015	Ezen-Can, A.; Boyer, K. E. A. [57]
AI	Q&A Systems	Educational Question Answering Motivated by Question-Specific Concept Maps	2015	Atapattu, T.; Falkner, K.; Falkner, N. [58]
AI	Q&A Systems (peer learning)	Reciprocal Content Recommendation for Peer Learning Study Sessions	2018	Potts, B. A.; Khosravi, H.; Reidsema, C. [59]
DS	Learning Analytics (the IoT can produce enormous benefits for society, such as advances in education)	The IoT for smart sustainable cities of the future: An analytical framework for sensor-based big data applications for environmental sustainability	2018	Bibri, S. E. [60]
DS	Behavioral analytics	The use of Big Data: benefits, risks, and differential pricing issues	2016	Simpson, D. [61]
DS	The emergence of Big Data provides excellent opportunities to change educational systems and programs (Predictions/innovation in education)	Evolution of big-data-enhanced higher education systems	2016	Li, S.; Ni, J. [62]
DS	Behavioral analytics (It allows the implementation of new approaches for the generation and analysis of data)	Big Data, new epistemologies and paradigm shifts	2014	Kitchin, R. [63]
DS	Personalized learning	Big Data and Its Research Implications for Higher Education: Cases from UK Higher Education Institutions	2015	Ong, V. K. [64]
DS	Learning Analytics	Una Revisión de la Literatura. Educación y Educadores.	2017	Rojas Castro, P. [65]
DS	Learning Analytics	La narración del aprendizaje a través de los datos	2017	Amo, D.; Santiago, R. [66]
DS	The authors found four opportunities using Big Data that can have a significant impact on education: (1) prediction of school performance; (2) recommendation system based on such prediction; (3) data-based Learning Analytics; and (4) personalized learning	Big education: Opportunities for big data analytics	2015	Cen, L.; Ruta, D.; Ng, J. [67]
DS	Learning Analytics	Learning Analytics Solution for Reducing Learners’ Course Failure Rate	2016	Govindarajan, K.; Kumar, V. S.; Boulanger, D.; Kinshuk [68]
DS	Personalized learning (applying Big Data to the educational cloud (EduCloud) platform suitable for teachers and students in Taiwan)	Taiwan digital learning initiative and big data analytics in education cloud	2016	Yang, S. J. H.; Huang, C. S. J. [69]
DS	Personalized learning/collaborative learning	Early-Stage Engagement: Applying Big Data Analytics on Collaborative Learning Environment for Measuring Learners’ Engagement Rate	2017	Lu, O. H. T.; Huang, A. Y. Q.; Huang, J. C. H.; Huang, C. S. J.; Yang, S. J. H. [70]
DS	Predictions/ Behavioral analytics	Typical Applications of Big Data in Education	2015	Yu, X.; Wu, S. [71]
DS	Recommendation Systems/ Learning Analytics	Online education big data platform	2016	Zhang, G.; Yang, Y.; Zhai, X.; Yao, Q.; Wang, J. [72]

shows the most representative documents that were studied in this research. The table indicates which computing discipline (ID: AI = Artificial Intelligence; DS = Data Science) the studied literature was classified as, and the literature was also specified according to its main usage of AI, such as adaptive learning, Q&A Systems, Intelligent Tutoring Systems, and Tutorial Dialog Systems. On the other hand, the main uses for DS were Learning Analytics (the IoT can produce enormous benefits for society, such as advances in education), behavioral analytics, predictions/innovation in education, personalized learning, collaborative learning, and Recommendation Systems. Likewise, the table contains information such as the title of the article or book chapter, the year of publication, and the author or authors.

Innovation in the educational field has focused on developing better learning processes that include activities that required a vast investment of time and distracted from the main objective pursued in the educational processes in the past. Some aspects in which Artificial Intelligence has contributed and will continue to contribute are the following: (a) personalized education based on the generation of student academic records and profiles, (b) intelligent analysis for the planning and design of innovative digital strategies and skills, (c) the interaction between humans and Information Technologies (IT) through new forms of communication (visual, audio–visual, listening, written, and gesticulation recognition, among others), and (d) a reduction in the intervention of human factors in iterative processes that do not require semiotic processes. The basic activities that can be carried out through the application of AI-based technology are the use of management chatbots and attention to students, the optimization of content searches to speed up and give certainty to the clarity of information and the transfer of knowledge, the evaluation of a large number of students in a short amount of time, the review of non-numerical tests related to writing and spelling, the use of collaborative learning environments in which several members participate to solve different cases, and the elaboration of diagnoses and control of academic profiles and of the students in an educational institution.

On the other hand, DS in the educational field has allowed for the discovery of new patterns, the prediction of the behavior of a system from the construction of models, the categorization of objects, and the validation of scientific theories, among many other actions related to data. The main benefits of the application of DS at present can be summarized in specific objectives such as prediction, intervention, recommendations, personalization, reflection, and iteration. Specifically speaking of the academic environment, results have been observed mostly in the prediction of academic performance, the personalization of teaching, the formation of working teams through the identification of personal patterns in individuals, the modeling of peer evaluation, and self-appraisal.

At present, all these benefits have been possible and have already benefited populations in the world. Specifically, in the period of confinement due to COVID 19, it was demonstrated that it is necessary to use the available tools and that we are on our way to be prepared for the transition to the era of the knowledge society.

3.5. Research Trends for DS and Education vs. AI and Education

Google trends has been used to predict trend directions in different areas. In this case, we used this tool to observe the interest shown in these types of topics among scientific groups. Many researchers focus on AI in education, and others focus on DS in education. In this paper, we present an analysis of some of the contributions found in the literature, and the search trends can be observed in Figure 2. The interest in researching topics related to AI in comparison to DS has grown from the year 2017 to now.

Figure 3 presents the search trends of topics related to “DS and COVID” and “AI and COVID” between 1 January 2020 and 31 December 2021.

Figure 4 shows the increasing interest in all these topics around the world. Two of the top countries interested in searching for these topics were the United States and India according to Google trends.

4. Discussion

Advances in digital technologies are constantly evolving and offer a lot of opportunities, which cause education to evolve by providing different tools aimed at improving the educational processes. These tools change teaching–learning methods and adapt to the needs of students, teachers, managers, and all of those involved in the teaching–learning process. Of course, these tools will evolve, and although they will perhaps be easier at the user level, the technology will be more complex. Disciplines such as AI and DS with their respective subdisciplines will be integrated into new platforms that are being used in the educational environment [73,74,75].

To name a few, there have been developments in virtual reality, augmented reality, virtual assistants, smart assistants, intelligent tutors, and automated learning management systems; education programs are needed at different levels in order to develop talent in people so that they can be prepared for developments in data analytics that help increase the competitive edge and sustainable growth of any organization [76]. These new programs should provide in-depth knowledge on the development of new materials for more efficient energy conversion systems and devices [77]. The knowledge, application, and use of these disciplines in education will require students, teachers, and managers to have the knowledge, skills, and attitudes necessary to successfully exploit all the resources provided. Therefore, teachers must be updated on pedagogical methods and on the use of ICT to minimize the digital divide since these are fundamental tools in the emerging models in the knowledge society [78,79], which has emerged from the pandemic era.

In this article, we presented a series of usages of AI and DS in education; this field is quite extensive, and other areas of applied research have yet to be analyzed, such as speech recognition for automatic dialogue systems, intelligent agents, predictive models, automatic evaluation, Recommendation Systems, gamification, and artificial vision systems, among others. On the other hand, almost all the analyzed Big Data and Learning Analytics use log files and do not give details on the type of unstructured data that they store. All the above technologies need to be used in order to transform education, and thus they include the concepts of sustainability and the vision of a better world. Sustainability is a way of thinking about the future in which environmental, societal, and economic areas are balanced in the pursuit of an improved lifestyle [80]. The massive use of natural resources, waste generation, and ecological damage are the reasons for such a dire situation. Big data analytics influence all areas that rely on technology [81,82,83]. Technology, the economy, and education are moving the gears to sustainable development in the knowledge society [84,85,86].

5. Conclusions

Even though AI and DS have been shown to help education systems use data to improve educational equity and quality in the developing world, there are still challenges to overcome in the near future. For example, public policies on AI and DS need to be established so that individuals can work collaboratively and make partnerships on national and international levels. In addition, the least developed counties are at the highest risk of not being able to ensure inclusion and equity due to economic, technological, and social divides. Another important situation that has to be considered for the implementation of AI and DS technologies in education to be successful is the preparation of teachers for these new educational trends. On the other hand, developers need to understand the best pedagogical practices to create solutions that fit the real context problems in classrooms. Inclusive and quality data systems are needed. Data collection and systematization have to be performed with proper processes to obtain quality systems. More significant research on AI and DS in education should be conducted in the coming years. It is necessary to continue to discuss the positive proposals and possibilities as well as the risks and possible solutions to develop better educational environments that move society towards sustainable social development; this will allow society to face new challenges, specifically those that occurred after the beginning of this pandemic period and those that will occur during a new era.