Collaborative learning and skill development for educational growth of artificial intelligence: A systematic review

Collaborative learning and skill development for educational growth of artificial intelligence: A systematic review. Contemporary Educational Technology


INTRODUCTION
The current revolution brought about by the emergence of artificial intelligence (AI) is transforming the way people live and interact with technology (Makridakis, 2017).Some of the features of AI, such as machine

Review Article
Mena-Guacas et al.

/ 17
Contemporary Educational Technology, 15(3), ep428 learning, have the potential to completely change many fields of knowledge as we understand them today (Wang & Siau, 2019).AI is being used to develop intelligent systems that can operate autonomously, improve decision-making processes and even provide human-like capabilities, such as natural language processing.
These particulars impact how education is delivered and received today enormous (Pereira et al., 2022;Celik et al., 2022).To positive use, AI-based tools and systems can personalize learning experiences, improve teacher productivity and enhance student attention.Virtual tutors could even be used to offer personalized attention to students in need and provide them with real-time feedback and support.However, while the benefits of AI in education may be numerous, it is crucial to approach this integration with caution and consider the ethical implications (Coeckelbergh, 2020).Furthermore, it is essential to ensure that AI systems are designed to augment and support human teachers rather than replace them.
In this context, we could think about the possibilities of AI in collaborative learning (CL) and competency development, the basis of contemporary pedagogical approaches.CL allows students to work together to explore and understand concepts, enabling a more profound understanding than can be achieved with traditional teaching methods.It focuses on the role of social interaction and teamwork in learning, where students work together to achieve common goals (Major, 2020).It fosters the development of essential skills such as communication, critical thinking, and problem-solving (Ghavifekr, 2020).
The development of competencies is also likewise determinant (Ferreira, 2013), as it allows students to become more proficient in understanding and applying the topics covered in class.CL and skill development can be enhanced through technology, such as online forums, blogs, and other web-based tools (Molinillo et al., 2018).They enhance blended learning, gamification, and project-based learning, so relevant in new pedagogies.They also develop relevant 21st-century skills, such as digital literacy.And among all these technologies, AI can undoubtedly stand out with the possibilities we have mentioned.
To improve the integration of AI in education, it would be crucial to establish the connections between CL and the development of competencies.Once configured, it would be necessary to specify which types of AI would be the most appropriate for its development.The complexity of today's organizational databases highlights the importance of complex technical skills as well as soft skills (Mitchell & Vaughan, 2022) and CL improved students' readiness for the workforce (Cioc et al., 2022) and transferable skills (Larraz et al., 2017), including improving one's ability to work with other students on assignments, listening to the ideas of others, learning from the ideas of others, and evaluating the ideas of others (Cioc et al., 2022).The skills or competencies required may be generic or specific.On the one hand, to face the challenges demanded by a changing and uncertain world, students need to develop generic competencies during their studies (Mykytiuk et al., 2022).On the other hand, today's companies require a workforce with specific competencies to perform increased complexity tasks (Marchi et al., 2019).
There are several ways in which skills (or competencies) development can be strengthened, e.g., incorporating interactive digital learning objects (McGuinness & Fulton, 2019), the use of small-space games (Bennett et al., 2018), behavioral skills training (Kirkpatrick et al., 2019), among others.CL is also one form, evidenced by the consistency with which studies have found evidence that peer interactions contribute to student achievement outcomes (Shafipour et al., 2018).
When collaboration manages to consolidate, educational action is more enjoyable and effective (Lee et al., 2011), individual learning outcomes are increased, and student satisfaction is better (Kwon et al., 2014, Kurucay & Inan, 2017;Hernández-Sellés, 2021).In addition, CL fosters students' sense of community, belonging, and influence (Luo et al., 2017) and contributes to maintaining emotional support (Hernández-Sellés et al., 2019), and is an indicator of student progress and student groups (Yucel & Usluel, 2016).Both the interaction between students and between them and the teacher are relevant to the collaborative process (Hernández-Sellés et al., 2019).However, each contributes differently: collaboration among students is fundamental for knowledge-building and competency development (Pulgar et al., 2022), while teacherstudent interaction positively influences students' active learning (Molinillo et al., 2018).However, CL has not been able to accurately master the information (Rui & Ying-Yan, 2012) generated in the interaction between actors.This is one of the reasons why one of the seven subjects most outstanding in AI in education is CL (Feng & Law, 2021).Also, computer-supported collaborative learning (CSCL) has gotten much attention in AI (UI Haq et al., 2020).This interest has even led to the proposal of a new conceptual framework called "intelligent tutoring supported collaborative learning (ITSCL)" (UI Haq et al., 2020).In the studies that have been done, teachers found the AI assistant helpful in collaborative learning environments because this technology can support the actions of teachers who implement the CL strategy, helping to detect students with needs and providing suggestions for intervention (Kasepalu et al., 2022).
AI has permeated all aspects of the educational field (Suning & Zhang, 2020), but the diversity of topics makes it difficult to address them all in depth, so the efforts are not enough development should focus on those lines that could generate more significant contribution.The analysis of the development of competencies through the promotion of collaboration should be one of them because competencies are the way to validate that the educational exercise has been successful and because collaboration is one of the most effective strategies to improve performance outcomes.Considering the above, two questions are proposed in this systematic review: (1) What are the characteristics of the studies that have been carried out about the relationship between skills development and CL? and (2) What should AI focus on to contribute to the understanding of competence development through CL?

Methodology
A systematic review was carried out in SCOPUS, and the PRISMA methodology was used.The search was done by crossing two of the three concepts (CL and development of competencies), which are the most frequently found in the available research, thus making it possible to answer the first question posed.The search algorithm is shown in Table 1.The concept of AI was not used in the search because it was the criterion for analyzing the documents.This issue is further elaborated in below.

Inclusion and Exclusion Criteria
Inclusion and exclusion criteria were defined to identify manuscripts to be included in the review.

Inclusion criteria
Mention or describe any relationship between interaction/collaboration and the development of some competence (for example, general, transversal, and specific, but it does not matter if classification is different).

Exclusion criteria
1.The interaction is not between people.
2. The concept of competence is associated with the dispute, contest, or competition and not with expertise, aptitude, skill, or suitability.
3. It is not possible to access the complete document.

Selection of Papers
With the search code, a total of 1,233 articles were found; after reading the title and considering the inclusion and exclusion criteria, 981 articles were eliminated.The remaining 252 were processed by reading the abstract, and 219 were eliminated.Three papers were discarded because the complete document was not available.30 works were included for the analysis by full screening test (Figure 1).

Main Findings
The evolution of the number of publications about skills development through collaboration is positive and will grow exponentially in 2022 (Figure 2).Four of the eighteen articles published between 2018 and 2022 explicitly mention COVID-19.
Table 3 shows information about the subjects or disciplines in which CL is applied to develop competencies.

Regarding ways of interaction
25 papers mention the relationship between the interaction and the development of competencies or skills directly, 27 papers indicate that there is a positive correlation between CL and skill development, 19 exhibits a face-to-face (FTF) interaction, nine computer-mediated interaction (CMI), one does not mention (NM), and one presents both interactions FTF and CMI (Figure 3).Over time, the interaction that stands out the most is face-to-face (Figure 4).

Table 2 (Continued). Papers information Title
Author(s) Year Effects of perceptions of the learning environment and approaches to learning on Chinese undergraduates' learning Guo et al. 2017 Tensions in mentoring medical students toward self-directed and reflective learning in a longitudinal portfolio-based mentoring system-An activity theory analysis  Mena-Guacas et al.

Regarding the skills studied in the papers in contrast to collaborative learning
Table 4 presents the list and the studies related to each.Some arouse more interest than others: cognitive, communicative, linguistics, collaborative work, teamwork, generic, and soft skills.It is essential to mention that 27 articles included in the systematic review found a positive relationship between interaction and the development of competencies.
Figure 5 shows the skills influenced by collaboration mentioned in papers over the last ten years.The slope of the curve is increasing.
To conclude the topic of developing competencies through collaboration, Figure 8 shows the studies that encourage interaction through teamwork.

Regarding the second question of this systematic review
The analysis of the 30 selected articles identified three categories that deserve attention: the objects of study, how the results are analyzed, and the types of AI that could be used.These categories were proposed by Chee et al. (2022).

How Results are Analyzed
Figure 9 shows the percentage, and Figure 10 the evolution over time, or use of the three forms of analysis, which can be: It is proposed as another category because the analysis of documents (texts, videos, etc.) is a highly used technique in AI.
The collective performance and learning content are subcategories defined by Chee et al. (2022).The category of individual performance emerged in the development of this systematic review.

Study Objects
In addition to skills, the most important object of study is related to the characteristics of learning (Figure 11), which has to do with academic performance, interest, and motivation, among others; moreover, interest in this topic has been growing since 2013 (Figure 12).

Types of Artificial Intelligence That Could be Used
Most of the studies are descriptive (Figure 13), and this is also the type that has been consolidated up to the year 2022 (Figure 14).The IA should support this type of study.

DISCUSSION
The first hypothesis about the cause of the growth in publications about the development of skills through collaboration (Figure 2) is the COVID-19 pandemic because it prompted many people to learn about the possibilities of collaboration in digital environments.Aguilar and Perez (2021) point out that CL is essential to a new educational approach characterized by social isolation and virtual learning experiences because of COVID-19.Other studies, such as Almusharraf and Bailey (2021), Dewi and Muhid (2021), Handini et al. (2022), Mustakim et al. (2021), andSafta-Zecheria et al. (2021), and also highlighted the importance of CL during the pandemic and the positive results obtained with it.The data from this systematic review suggest that this may not be the final answer since many of the papers reviewed present experiences before the pandemic or face-to-face interactions (Figure 2), and only 22% of the papers published between 2020 and 2022 explicitly mention the pandemic.The rest may have taken the data earlier or do not expressly mention it as an element they consider relevant to the research, which shows that interest in the topic is not only a result of the pandemic.Moreover, it coincides with the findings of Feng and Law (2021), who mentioned that CL is one of the most studied education topics with AI.
Although there is a tendency to develop isolated studies by discipline (García-Chitiva, 2021), research on skills development through collaboration is done in almost all of them (Table 3).This shows that the interest in developing skills supported by collaboration does not depend on discipline.
Only one article combines face-to-face and computer-mediated interaction (Figure 3).There is still a predilection for face-to-face interaction, even though teaching in 21st-century higher education is moving towards e-Learning or b-Learning teaching models (Sáiz-Manzanares et al., 2021).This may happen because educational institutions have been slowly moving to transform their offerings towards e-learning and blearning.The COVID-19 pandemic has changed the face of higher education, many universities have adopted the hybrid or blended medium of instruction (Singh et al., 2021), so the modality transformation could be much more accelerated in the coming years.
Research on the correlation between collaboration and skills development has aroused more significant interest in studies conducted in education, with a powerful tendency to increase over time.For example, in 2013, articles only mention the influence of interaction in two competencies (linguistic and cognitive), while in 2022, its influence is noted in eleven (generic, specific, teamwork, sustainability, relational, technical, linguistic, professional, structural tasks, collaborative work and communicative) (Figure 5).This shows that collaboration has been found to contribute to more competencies than initially thought.Several studies in different fields of learning, such as those by Gómez-Galán (2020), Hei et al. (2020), Herrera-Pavo (2021), and López-Meneses (2020), and have presented the reality of this fact.
The type of interaction most often studied is between students (Figure 6), indicating that teachers and tutors recognize that students have managed to self-regulate and self-manage their learning.In addition, student-student relationships can play a role in contributing to transformative learning and lead to personal growth (Kelleher, 2020) and is one of the aspects that allow for a harmonious classroom environment (Luo et al., 2022).This shows that the decision to prioritize the study of student-student interaction is a wise one.
Regarding the second question proposed for this review, AI algorithms should concentrate efforts on analyzing the entire group (Figure 9), but in the educational exercise, it is essential to have individual data for decision-making and strategy design.In a similar way, although in other technological contexts, authors such as Pastora and Fuentes (2021) and Webber and Zheng (2020) have an impact on this issue.This is reinforced by what is seen in 2022, as the interest in individual performance equaled the collective (Figure 10).This shows that the potential use of algorithms will be more significant as they provide data about the development of competencies at the individual and group levels.This does not imply that the indicators or analyses are the same at both levels, but both must be carried out.For example, at the individual level, the processing of the content of a student's document would be automated with AI to identify the development of competency and cross-referenced with the quality of their interaction, while at the group level, group patterns would be placed.Sentiment analysis has not been used much in the study of competence development through collaboration (Figure 11), but it has been explored in depth in the AI field.Such an advance would make a significant contribution in this field because positive feelings favor interaction, promoting collaboration and skills development.This has been pointed out in several studies, e.g., (Zhipeng et al., 2022) pointed out that to achieve satisfactory learning outcomes, students must maintain positive emotions during the CL process, and Dehbozorgi et al. (2021) said that there is a strong positive correlation between students' positive emotions while interacting with each other with their performance in the course.Sentiments, interaction, collaboration, and competencies are four elements connected, so it is necessary to analyze them together, and AI can support this task.The line of work could be to explore the participants' feelings and the interaction that occurs in learning environments; then, it would be necessary to assess the collaboration achieved between students and teachers through how they interact and finally to evaluate the development of competencies.Discourse patterns are another advanced topic in AI, as discussed by Chee et al. (2022) and Dowell et al. (2019) but developing competencies through collaboration is not yet sufficiently addressed.
From another perspective, the results show that descriptive studies abound (Figure 13 and Figure 14), but those that provide decision-making inputs (predictive and diagnostic) are almost not carried out.In contrast, the research in AI-assisted decision-making is experiencing tremendous growth (Schemmer et al., 2022) because it provides insights and methods that enable decision-making (Yong et al., 2022).Consider CL as a practical resource for developing competencies with AI, for example, defining the most effective way to distribute learners in groups for all participants and assessing the quality of interaction in learning environments, among others.

CONCLUSIONS
Conclusions are proposed based on the two questions formulated for this review: Regarding the first one, it is essential to mention that studies on developing competencies through collaboration are rising.The main characteristics of the studies are that they contribute to different areas or disciplines and prioritize face-to-face interaction, and since 2019 there has been an increasing interest in computer-mediated interaction.The most studied skills are cognitive, communicative, collaborative, and generic or soft skills.Finally, the interaction between students is the one that arouses the most interest among researchers.About the second question, AI development efforts could focus on algorithms that enable collective and individual performance analysis, boost the study of sentiment, and provide tools for diagnostic and predictive analysis that focus on providing information for decision-making.AI has become an increasingly popular topic in recent years, and interest in investigating its potential educational applications has increased, but much remains to be explored in this area.In addition, although it has great potential to improve education, it should be approached with caution.More research is needed to fully understand its impact and how best to implement these technologies in the classroom, minimizing its drawbacks, which may be relevant, and making a truly effective and productive use of it.

Figure 2 .
Figure 2. Number of papers per year (Source: Authors)

Figure 3 .Figure 4 .
Figure 3. Percentage of items according to interaction type (Source: Authors)

Figure 10 .Figure 11 .
Figure 10.How results are analyzed in the papers by year (Source: Authors)

Figure 12 .Figure 13 .Figure 14 .
Figure 12.Topics of study, according to how the results were analyzed (Source: Authors) (a) What are the characteristics of the studies that have been carried out about the relationship between skills development and CL and (b) What should AI focus on to contribute to the understanding of competence development through CL?

Table 3 .
Subjects or disciplines