With the rapidly increased use of digital technologies in education due to the COVID-19 pandemic, it is important to discuss these technologies' impact on the teaching profession. Using thematic analysis and epistemic network analysis (ENA), we analyzed 114 social media posts by Finnish education professionals commenting on an opinion piece about technology partly taking responsibilities usually attributed to a teacher. Out of the analyzed posts, 32 were supportive, 30 ambivalent, and 52 critical towards the presented scenario. The epistemic network analysis graphs showed some diferences between posts with a diferent attitude. Supportive posts, on average emphasized technological capabilities and their connections with teacher workload and self-directed/self-regulated learning. In comparison, the critical posts on average emphasized human presence and its connections with pupil diversity and technological capability. Our findings both reveal the relevant themes in the discussion about technologies' impact on the teaching profession and underline the diferences in supportive and critical argumentation.
While digitalization and automation have disrupted many industries in the last few decades, education has largely been an exception to this trend. As Selwyn puts it, “most people intuitively feel that education is an essentially human undertaking” and “the belief persists that learning is something best guided by expert human teachers in socially rich setting” [1, p. 1].
However, now that the COVID-19 pandemic has interrupted classroom learning for at least 9
out of 10 students worldwide [
In the present study, we examine Finnish education professionals’ thoughts on digital technologies in education and their impact on the teaching profession. First, we present relevant background related to artificial intelligence in education and technology-enhanced self-regulation of learning, followed by our research questions.
Novel technologies such as machine learning and artificial intelligence are changing labour
markets and disrupting industries. Computerization of work is moving fast from routine tasks
involving explicit rule-based activities to non-routine cognitive tasks [
Regarding the teaching profession, teachers are considered to have a low risk of
computerization [
It is essential to acknowledge, that no technological innovation will inevitably change teaching
profession. As Selwyn [
Self-regulated learning refers to how learners systematically activate and sustain their cognitions, motivations, behaviors, and afects toward the attainment of their learning goals [ 13]. In a classroom setting, the teacher can support learners’ self-regulation and then gradually decrease the amount of support, promoting the learning of self-regulation skills.
With the development of adaptive learning technologies (ALTs, see e.g., [10]), this role of scafolding support may partly be transferred to technological systems. Researchers have envisioned “a hybrid human-system regulation with fluctuating boundaries between student and system control” where “embedded learning analytics techniques can support a new generation of SRL support that adjusts external regulation based on insights gained from data” [10, p. 473] and presented a framework for human–AI hybrid adaptivity [14], describing how humans and educational AI systems can augment each other in many ways. Moreover, it has been pointed out that the development of human-AI systems should engage practitioners (in this case, teachers) throughout the lifecycle of system development and that these systems should be studied in real-world contexts [15]. While these technologies are still at their preliminary stages, their sophistication will grow over time.
In summary, it is highly likely that new technologies will change the teaching profession in one way or another. Therefore discussing these topics within the community of education professionals is extremely important.
With the massive increase in the use of digital tools in teaching and learning during the pandemic, it is important to investigate implications to educational arrangements in general and to teaching profession in particular. Based on literature, we know that machines can successfully handle many routine teacher tasks such as grading. However, supporting learners’ self-regulation has been an exclusively human endeavour and central to the teaching profession. As digital systems get more sophisticated, it is important to discuss the possibilities and limitations of a machine to guide, encourage and support students with their self-regulation broadly in the educational community.
In the present study, we analyze social media posts commenting on an opinion piece published in the largest daily newspaper in Finland. The author of the opinion piece describes how learning of new content is increasingly guided by adaptive learning technologies instead of a human teacher and how this kind of shift would free teachers’ time to focus on tasks where humans are best. Our research questions are following: • RQ1: What kind of argumentation Finnish education professionals use in social media discussions to support or criticize a scenario of adaptive learning technologies taking responsibilities traditionally attributed to teachers? • RQ2: How do Finnish education professionals see the role of a teacher in contrast to the role of technology in supporting pupils’ self-directed and self-regulated learning?
In Finland, all schools shifted to remote learning because of the covid-19 pandemic for eight weeks during spring 2020. Afterwards, schools have shifted between contact teaching and remote learning depending on the development of the epidemic in each region.
During the pandemic, there have been active discussions in Finnish media about learning with digital technologies. One contribution to this discussion is an opinion piece [16] written by a Helsinki-based elementary school teacher and published on April 19th 2021 by Helsingin Sanomat, the largest daily newspaper in Finland. The author suggested introducing weekly remote learning, where elementary school pupils would study remotely one day a week leveraging digital learning platforms. He describes how this kind of shift would free teachers’ time to focus on things where humans are best, such as contextual interaction.
The piece generated lots of comments in social media, especially in two Facebook groups: a general forum for discussion among Finnish education professionals (ca. 17 500 members) and a group specializing on the teaching profession (ca. 13 000 members). Comments were posted during April 2021.
Social media posts were gathered from the Finnish education professionals group (N=48), teaching profession group (N=81) on Facebook, Twitter (N=12), and newspaper comment board (N=10). Posts that did not include any argumentation (e.g. posts with only one word or emoticon) were removed from the dataset, ending up with 114 posts. In addition to the post content, the amount of social media reactions (i.e. likes) were collected for each post. The opinion piece as well as all the comments were originally written in Finnish and the excerpts presented here have been translated into English by the authors.
2.3.1. Attitude towards the scenario.
The posts were rated by both authors as supportive, critical or ambivalent, based on the attitude towards the presented scenario. The inter-rater reliability was moderate (Cohen’s kappa 0.68)[17].
Inductive thematic analysis was used to analyze the themes included in the social media posts. Both authors took part in the process. First, the first author coded all the posts and created the initial coding scheme. Typically each post was assigned two to five codes. Next, the second author used the coding scheme with 20 first posts. Then, inter-rater diferences were discussed, and the coding scheme was readjusted (e.g. two of the initial codes were removed). Finally, both authors used the agreed flat coding scheme with all the posts. The codes with their descriptions and inter-rater reliability are presented in Table 1. Inter-rater reliability ranged from moderate (Cohen’s kappa 0.61) to strong (Cohen’s kappa 0.84)[17].
In this study, we applied Epistemic Network Analysis [18, 19, 20] to our data using the ENA 1.7.0 Web Tool [21]. Our ENA model included the codes presented in Table 1. We defined conversations as all posts associated with a single social media forum. The ENA model normalized the networks for all units of analysis before they were subjected to a dimensional reduction, which accounts for the fact that diferent units of analysis may have diferent amounts of coded lines in the data. For the dimensional reduction, we used a singular value decomposition, which produces orthogonal dimensions that maximize the variance explained by each dimension (see [18] for details).
In this study, ENA was used to compare mean networks of posts with supportive, critical or ambivalent attitude towards the presented scenario. Networks were visualized using network graphs where nodes correspond to the codes, and edges reflect the relative frequency of cooccurrence between two codes. The positions of the network graph nodes are determined by an optimization routine that minimizes the diference between the plotted points and their corresponding network centroids. Because of this co-registration of network graphs and projected space, the positions of the network graph nodes and the connections they define can be used to interpret the dimensions of the projected space and explain the positions of plotted points in the space. Our model had co-registration correlations of 0.93 (Pearson) and 0.93 (Spearman) for the first dimension and co-registration correlations of 0.91 (Pearson) and 0.91 (Spearman) for the second. These measures indicate that there is a strong goodness of fit between the visualization and the original model.
The number of posts including each code as well as sum of likes of those posts are presented in Table 2. Out of 114 posts analyzed, 32 (28 %) were supportive towards the presented scenario, whereas 30 (26 %) were ambivalent and 52 (46 %) critical. While this presents the attitudes of professionals taking part in the discussion, the sum of social media likes shows attitudes of the larger community. Out of 1251 likes, 103 (8 %) were connected with posts showing supportive attitude, whereas 192 (15 %) were connected with posts showing ambivalent and 956 (76 %) with posts showing critical attitude.
The epistemic network analysis graphs are presented in Figure 1. The most central code was Educational arrangements (ARR), having strong connections with all other codes. When comparing the networks of supportive and critical posts, it is clear that diferent themes are ARR Educational arrangements PRE Human presence and interaction TEC Technological capability
WOR Teacher workload and efficiency DIV Pupil diversity and equality
SEL Self-directed/self-regulated learning All posts Ambivalent Supportive Critical highlighted. In the supportive network, the focus is on connections with Technological capability (TEC) and in the critical network on connections with Human presence and interaction (PRE)
Moreover, it is interesting to compare which connections are missing in each graph. In the supportive network, connection between Pupil diversity and equality (DIV) and Human presence and interaction (PRE) is missing, as well as connection between Technological capability (TEC) and Human presence and interaction (PRE). Respectively in the critical network, the connection between Technological capability (TEC) and Teacher workload and eficiency (WOR) is missing, as well as the connection between Technological capability (TEC) and Self-directed/self-regulated learning (SEL).
In the following, we take a deeper look into the roles of technology and teacher in guiding and supporting pupils. 3.2. The Roles of the Teacher and Technology in Supporting Pupils Many comments emphasized pupils limited self-directed and self-regulated learning skills. This was often coupled with the notion of challenges in supporting self-regulation remotely: “Uppermost the delusion that you could just ’leave’ the children to study the vocabulary etc. Not at all. Online the teacher presence is highlighted more intensively so that high-quality learning and continuous assessment can happen. It requires a lot: both from teachers and pupils. The most self-directed learners can surely do something as described, but I would not in any case - and fortunately it’s not legally possible categorically increase remote learning in basic education.”
“[–] The pupils aren’t self-directed except for a few cases. Most of them need interaction and discussion about other things among teaching. That works best live, worse remotely.”
Some of the critical comments directly addressed the shortcomings of technology in guiding and supporting pupils:
“This digital-environment-self-directed-learning mantra has been tooted as forthcoming for thirty years now. Nevertheless, give any ’self-directing’ learning material or digital platform to a group of 24 pupils and follow the situation for 45 minutes. I’ll eat my hat if even half of the group is doing what they’re supposed to with the program, or if they’re even in the program. In small bits, yeah, but replacing teachers, doubt it.”
Other comments did not mention the technology but emphasized that guiding and supporting should be carried out by the teacher:
“Oh my god! Teacher’s role is to follow how learning proceeds and help when necessary. For example, there are a lot of tools and collaborative learning plans for multiplication tables.”
On the other hand, there were many comments which described how technology would be able to support pupils :
“The technology should distribute exercises, grade them, correct errors, praise success, and guide otherwise too. The teacher would get some kind of overview.”
“Using digital materials does not necessarily lead to an increase in distance education, but creates also possibilities for individual work by providing direct feedback and diferentiation possibilities. At school, or at home.”
As described in the Introduction, a skilled teacher gradually decreases the amount of support for learners to learn to regulate themselves. It is an interesting question, whether this kind of ifne-grained adaptive regulation support is expected from a machine. One comment included the idea about developing students self-directed and self-regulated learning skills, but it was left a bit unclear whether this development is seen as a side efect of individual studying with technology or as a consequence of scafolding by either teacher or technology: “I think that while provocatively formulated, the idea is rather good. Many digital tools fit really well to diferentiation and e.g., vocabulary drilling. Pupils could and should be guided to plan, test, and diferentiate their learning themselves and, for example, proceed to more advanced exercises or broader vocabulary when the basics work out. [–]”
In another comment, it was suggested that self-regulatory skills could be practised with technology, but there should be a human “backup”:
“[–] it’s worthwhile to let those with [self-regulatory] problems to have a chance to practise at school. In short snippets and so that someone is there to return drifters back to the track.”
There were no comments about technology being capable of directly helping with challenges related to pupil diversity and equality. One comment even pointed out the contrary: “Technology can help a lot in teaching, but will never be a substitute for teacher’s help to those with challenges in studying - and it is an increasing crowd!”
However, some comments noted that technology could indirectly help by freeing up teachers’ time:
“This can also be seen in a way that if those competent in the independent study are in remote learning, those who are unable may get tailored contact teaching.” “Let’s teach just those who need teaching. Others might find learning even easier alone on their own.”
In summary, in the posts that were supportive towards the vision presented in the opinion piece, technology was seen as capable of guiding and supporting students with their selfregulation. On the other hand, the critical posts mostly, rather than criticizing the technology, emphasized the role of the teacher. It seems that the possibilities of technology in guiding and supporting students are recognized to some extent, but technological scafolding alone is not considered suficient. Therefore, the role of teacher is seen as necessary even with adaptive learning technologies available.
In line with previous work among diferent industries [
In the supportive posts, on the other hand, 66 % mentioned technological capability. Even these posts described the technological possibilities on a rather conservative level, at least if compared to visions of hybrid human-system regulation [10]. For example, the possibilities of learning analytics were mentioned only once.
Moreover, while ethical and legal issues related to collecting and using student data in various applications are active topics in the learning analytics community, these themes were largely absent in this discussion among education professionals.
The present study has some limitations. First, the sample size is small with only 114 social media posts. This is due to the fact that posts considering such a specific topic are scarce - we were lucky to find this kind of discussion occurring spontaneously. In future work, alternative data gathering methods such as surveys and interviews should be considered.
Second, a clear limitation in this kind of research setting is the question of representativeness. We cannot objectively tell, how well the views presented in the analyzed posts represent the views of Finnish education professionals in general. However, the Facebook groups in question are rather large when taking into account the population of Finland, so at least a substantial share of Finnish education professionals had a chance to take part in this discussion.
We analyzed 114 social media posts commenting on an opinion piece about technology partly taking responsibilities usually attributed to a teacher. Our findings indicate that Finnish education professionals mostly do not see adaptive learning technologies disrupting the teaching profession.
There were some diferences between supportive and critical posts. Supportive posts connected technological capabilities and self-directed or self-regulated learning, emphasizing that technology can also guide and support students. Critical posts connected human presence, educational arrangements, and student diversity and equality, emphasizing the importance of teacher’s presence in addressing pupils’ varying needs.
Our findings reveal themes relevant when discussing the development of adaptive learning technologies and their potential impact on teaching profession. Moreover, the findings increase the understanding of how supportive and critical argumentation on technology difer from each other. teaching augmentation for k-12 classrooms, in: Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, 2020, pp. 1–17. [8] O. Viberg, M. Khalil, M. Baars, Self-regulated learning and learning analytics in online learning environments: a review of empirical research, in: Proceedings of the tenth international conference on learning analytics & knowledge, 2020, pp. 524–533. [9] E. Mousavinasab, N. Zarifsanaiey, S. R. Niakan Kalhori, M. Rakhshan, L. Keikha, M. Ghazi Saeedi, Intelligent tutoring systems: a systematic review of characteristics, applications, and evaluation methods, Interactive Learning Environments 29 (2021) 142–163. [10] I. Molenaar, A. Horvers, R. S. Baker, Towards hybrid human-system regulation: Understanding children’srl support needs in blended classrooms, in: Proceedings of the 9th International Conference on Learning Analytics & Knowledge, 2019, pp. 471–480. [11] R. Winkler, S. Hobert, A. Salovaara, M. Söllner, J. M. Leimeister, Sara, the lecturer: Improving learning in online education with a scafolding-based conversational agent, in: Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, 2020, pp. 1–14. [12] J. A. Pesonen, ’are you ok?’ students’ trust in a chatbot providing support opportunities, in: P. Zaphiris, A. Ioannou (Eds.), Learning and Collaboration Technologies: Games and Virtual Environments for Learning: 8th International Conference, LCT 2021, Held as Part of the 23rd HCI International Conference, HCII 2021, Springer, 2021. [13] D. H. Schunk, J. A. Greene, Historical, contemporary, and future perspectives on selfregulated learning and performance, Routledge, 2017. [14] K. Holstein, V. Aleven, N. Rummel, A conceptual framework for human–ai hybrid adaptivity in education, in: International Conference on Artificial Intelligence in Education, Springer, 2020, pp. 240–254. [15] K. Holstein, V. Aleven, Designing for human-ai complementarity in k-12 education, 2021.
a r X i v : 2 1 0 4 . 0 1 2 6 6 . [16] T. Luoto, Peruskoululaisten osa-aikainen etäopiskelu tehostaisi opettajien ajankäyttöä,
Helsingin Sanomat (2021). URL: https://www.hs.fi/mielipide/art-2000007928464.html. [17] M. L. McHugh, Interrater reliability: the kappa statistic, Biochemia medica 22 (2012) 276–282. [18] D. W. Shafer, W. Collier, A. R. Ruis, A tutorial on epistemic network analysis: Analyzing the structure of connections in cognitive, social, and interaction data, Journal of Learning Analytics 3 (2016) 9–45. [19] D. Shafer, A. Ruis, Epistemic network analysis: A worked example of theory-based learning analytics, Handbook of learning analytics (2017). [20] D. W. Shafer, Quantitative ethnography, Lulu. com, 2017. [21] C. L. Marquart, C. Hinojosa, Z. Swiecki, B. Eagan, D. W. Shafer, Epistemic network analysis (version 1.7.0), 2018. URL: http://app.epistemicnetwork.org.