Technology Enhanced Learning Overview

Technology innovation has been selected as case study is Technology Enhanced Learning (TEL). Following the description by (World Education Forum, 2015), TEL is the integration of digital technology to improve the learning and teaching processes by adding value to academic performance outcome of students. We can recognise TEL as a technology innovation in line with (Vincent-Lancrin et al., 2019) in terms of product innovation and process innovation. Accordingly, with (OECD/Eurostat, 2018) these two innovation categories are sometimes intertwined, in particular, in the context of digitalisation. Hence, it is possible to comprehend TEL as an innovation in the context of educational organisations by the introduction of a new product which could also prompts process innovation, e.g. e-learning; news ways of organising educational activities, e.g. monitor tools that facilitate the creation of long-term individual educational plans and new external techniques which could transform that the manner in which the schools, for example, communicate with students and engage with labour markets and communities. The choice of TEL as a technology innovation case-study dwells on the potential which digital technologies can bring into education. As mentioned by (Duval, Sharples and Sutherland, 2017) “TEL harnesses the power of interactivity and has the potential to enhance what is learned, how we learn and how we teach”. As has been mentioned by (Vincent-Lancrin et al., 2019), one of the main innovations in education in the previous decade has been the increasing use of information and communication technology (ICT) to search for ideas, information and complete tasks independently along with the introduction of new pedagogical innovations such as the flipped classroom. From this perspective, it is possible to see TEL not just as a product and process innovation but also as a social innovation in the way that addresses not just a market need but a social one also (Mulgan et al., 2007).As a service innovation, as (Barras, 1986) has explained, the schools currently experience the requirements of arranging for increasing utilisation of ICT within the classroom. This is congruent with the overall technological ethos to which the new generation is subjected.

Conole et al. (2008) also highlighted this problem, namely, ‘a mismatch between our current offerings and student use and a further mismatch between institutions’ perceptions of utilisation of technology by students and actual use’ (p. 519).Therefore, product innovation shadows the urgent requirement that schools currently experience to innovate the process for the proposed educational services which they offer. As a regional, national and sectorial innovation in the way that schools are connected through a network of institutions (Dodgson et al., 2008), which share between them common policies and standards. “Successful integration of ICT into teaching and learning requires rethinking the roles of teachers and reforming their preparation and professional development. It calls for promoting a culture of quality in all its aspects: staff support, student support, curricula design, course design, course delivery, strategic planning and development” (UNESCO, n.d). A point that is questioned is the innovative nature of TEL: disruptive, sustaining, or efficiency? Accordingly, with (Christensen 1997, p. xv), TEL can be related with all three categories; sustaining innovation are the ones which permit the improvement of something which we have been previously performing , while disruptive technologies are the ones which prompt new practices. Christensen, Bartman, and van Bever (2016) introduce efficiency innovations which reduce cost by eliminating labour or by redesigning products to eliminate components or replace them with cheaper alternatives, but, underlines that this approach can translate itself as “a race to the bottom”. From this perspective (Christensen et al., 2016) argue that TEL, as an efficiency innovation can be potentially pedagogically constrictive. From the perspective of open vs close innovation. As remarked previously, the introduction of TEL can be seen as an open innovation in the sense that combines internal needs and external factors which prompt the development of new technologies (Chesbrough et al., 2006).

The selected emergent technology is Artificial Intelligence (AI) in Education (AIED). AI, as a broader concept, is considered a disruptive technology, it aims to mimic the human cognitive behaviour (Touretzky et al., 2019). For example, when a machine is able to process and generate data in an “intelligent” manner, that is AI (Data Science Central, 2018). Machine-learning (ML) and Deep Learning (DL) are sub-branches of AI, with the ability to analyse statically automated and non-automated data and update their-decision makers algorithms so they are able to learn over time. ML makes use of neural networks in the same way as humans’ brains to process data through a series of layers of algorithms designed to recognise patterns and adapt to the output required (Investopedia, 2020). This means that for ML to be efficient, the bigger the data sample, the better. AI is already in place in some schools through the automation of administrative tasks, e.g. grading. Cram101 and Netex Learning are mechanisms which are utilised for data processing, e.g. a textbook, into smart content such as summaries, flashcards and practical tests (eLearning Industry, 2020). AI also addresses one of education major difficulties: Personalised Learning. Traditional education plays by the motto: “one size fits all”. This means that traditional education is designed to target as many pupils as possible by focusing on the 80% of the average. Which leaves the remaining top and bottom 10% struggling to achieve their potential or to keep up with the class. (Inc., 2020). AI enables instant feedback, creating tailored assignments and individual curriculums which bridge this learning gap. AI capacity to model complex data promises a breakthrough in terms of data-driven forecasting, paving the way for schools to better understand the process of human learning, creating adaptive learning techniques with customised tools which could support each pupil from Ks1 to higher education (eLearning Industry, 2020).

All this hype around AI, doesn’t come without flaws and space to improve though.

When handling personal data, schools have to embed their practice with several privacy and security mechanisms, such as Europe’s new General Data Protection Regulations (GDPR) (Inc., 2020). Some AI learning systems also possess the Assessment and Learning in Knowledge Space (ALEKS), a popular online learning platform that boosts and is able to give an accurate picture of pupils’ knowledge level and have been found to have discrepancies when applied in different social backgrounds (Lexalytics, 2020). Different AI companies have different approaches to what is the role of AI in education. Having in one side, AI assisted where teachers keep an active role with creative tasks. On the other side, in AI lead technology, teachers are passive actors just intervening when there is a problem and AI takes over the teaching and learning process (Lexalytics, 2020). Some studies also found that automated essay-scoring systems can be easily fooled with gibberish and grade minorities differently (Lexalytics, 2020). This means that AI is susceptible to perpetuate social bias failing to deliver the same level of personalisation expected from a human teacher.

a) The road map format chosen is in line with the second perspective presented by (Phaal, Farrukh and Probert, 2004). That addresses the “environmental landscape, threats and opportunities for a particular group of stakeholders in a technology or application area”. When addressing the role of AI in education (AIED) and future possibilities, there was the need to consider what is understood by AI; what implications AI has had so far and will have in the future of Education and in what way the development of AIED has been an extension of the general development of AI, machine learning and deep learning, not just in education but across different sectors. The roadmap used in this work did not aim to focus on the development of AI as a specific technological artefact, rather as an innovation in the educational sector, contributing to massive changes in the pedagogical processes of teaching and learning (OU, 2020). The map’s time frame, focuses on the period between 1950 and 2036. The second half of the 20th century witnessed the birth of AI and an increasing growth of AIED, that is expected to increase at an accelerating pace (Holmes, Bialik and Fadel, 2019), has showed in the map. The time line frames key milestones that have already occurred during the development of the technology and that are expected to happen in a near future. AIED has not been following the typical market drivers, as expected in a for-profit company. State schools are non-profit, and pedagogical drivers are a better indicator of the development and application of AIED. Despite the increasing investment in AI in the last decade, the implementation of AI in education has been slow (‘ai100report10032016fnl_singles.pdf’, 2016). This can, in part be explained due to the lack of investment in the educational sector that does not bring the same profit returns in the short-term as other industries, such as technology & communication and automotive industries (price economics, 2020). Due to the lack of strong evidences that AI technology improves students’ learning achievements, such outcomes could take place. Throughout the research done, the map shows that the milestones don’t happen in a horizontal linear way. Like so, it is possible to state that there are three major breakthroughs (mechanical teaching machines; machine learning; deep learning) that are articulated between them through other milestones such as adaptive learning and search optimization. The map is then divided vertically in technological advancements, pedagogic drivers and challenges. Each of these vertical sections follows the horizontal non-linear sequence of the past and expected future progresses, using colour labels which match each corresponding milestone, framed within the time-line presented.

b) Has showed on the map, the first step towards AIED started with the work of the psychologists Sidney Pressey, Professor at Ohio State University 1920s, and B.F. Skinner, Professor Harvard Uni. 1948-1974, known as the father of Behaviourism. Sidney Pressey saw the mechanical teaching machine as a possibility to guarantee immediate feedback to students. His work focused on improving the multiple-choice test to consolidate students’ learning. Skinner’s teaching machine was an improvement comparing with Sidney’s machine. Skinner’s machine was able to distinguish between the subject content, pre-programmed into the machine and the student’s achievement. In this sense, Skinner machine foreshadowed AIED’s intelligent tutoring system and machine learning (Holmes, Bialik and Fadel, 2019). However responsive, in the sense that it would provide instant feedback by revealing the correct answer, Skinner’s machine, cannot be considered adaptive. This means that it would not adjust either the questions or the presentation sequence to the needs of the student (Holmes, Bialik and Fadel, 2019). During the fifties, Gordon Pask developed the first adaptive teaching machine. This key milestone, together with the advent of the first computers available, boosted the spreading of the first Computer-Aided Instruction (CAI) systems during the sixties and seventies. PLATO is considered one early influential CAI system, allowing a great number of students to access interactive teaching materials via remote controls (Holmes, Bialik and Fadel, 2019), which can be seen as an early example of the later Massive Online Open Courses (MOOCs). However, CAI systems lacked adaptability provided that the sequence of topics, the content provided and the response to students’ performance were predefined, missing students individual learning needs (Holmes, Bialik and Fadel, 2019). During the sixties, CAIs were followed by the TICCT (time-shared interactive computer-controlled information television) systems. Traditional CAI systems were succeeded by SCHOLAR. The first Intelligent Tutoring System (ITS) credited to Jaime Carbonell, 1970, was adaptive, tailoring individual responses to students’ statements by making use of a semantic network. In this sense, SCHOLAR is considered the first system to implement AI techniques (Holmes, Bialik and Fadel, 2019).

Throughout the Eighties, till the first decade of the 21st century, experts stated that AI had fallen within a declining period. As the map explains, this was due to the lack of computer power and access to large amounts of data (freecodecamp, 2020). AIED as we know it today has been boosted by the advent of DL. DL is an AI technology, conceptualised by Geoffrey Hinton in the Eighties,that uses an “algorithm structured similar to the organisation of neurons in the brain” (freecodecamp, 2020). During the second decade of the 21st century, the exponential growth of computer power and the access to large amounts of data created the perfect conditions to the field of learning analytics to sprout up (‘ai100report10032016fnl_singles.pdf’, 2016). A renewal of interest in the development of technologies which enhance teaching and learning had occurred and AIED technologies available nowadays appeared. This explosion of data also gave birth to new concerns regarding AI and ethics.

AI has been the driving technological pull of the first half of the 21st century (Figure 1).

As Figure 1 shows, the pace that AI/AIED has developed, during the last decade, has been increasing, and so, there is an element of difficulty in predicting what will be the distant future in a field moving so fast (Holmes et al. - 2019). In the next coming years, it is expected that DL will take over the AI industry. While traditional ML processes data in a linear way, DL hierarchical functions enable the machine to process large amounts of data in a non-linear manner. Making possible for AIED to progress from learning to labelled data to, finally, learning from reading and watching (Holmes et al. - 2019). Some very near future expected developments in AIED are: Machine Translation (MT) will make it easier to translate educational material into different languages; adaptive learning will facilitate the pedagogical individualisation required while the number of students per class increases; the line between formal and non-formal learning will fade away as the number of cloud computing learning platforms increases (‘ai100report10032016fnl_singles.pdf’, 2016); as it has been happening in the previous decades, regarding the shift towards a knowledge based economy, the knowledge value will decrease, while problem-solving, creativity, inter-personal, 21st century skills, value will increase (David Price, 2013). This means that AIED will reshape not just ‘how’ we teach but ‘what’ we will teach (Holmes et al. - 2019). One of the key issues regarding AIED in the present and near future is related with the access that schools have over large amounts of student’s personal data. This concern has been triggered in the last decade through the creation of security and privacy mechanisms such as the GDPR (Inc., 2020). On the same note, DL feeds on big data, but accuracy takes time to be learnt. As has been demonstrated on the map, it will be an international effort to make sure that AI-based decisions are devoid of biases, not allowing the spread of discrimination against race, sexual orientation or other factors (‘ai100report10032016fnl_singles.pdf’, 2016). Another question is, what will be the role of AIED? Some examples tend towards an AIED that leads fully to the teaching and learning process (Lexalytics, 2020).

Summarising, AIED has a great potential to enhance teaching and learning processes. However, this involves the need to understand better, how do people learn, how to apply this knowledge into the classroom and what technologies can facilitate this process? At the same time, it is also crucial to understand the potential dangers of using technology in a large scale for education purposes.

Bibliography

Duval, E., Sharples, M. and Sutherland, R. (eds) (2017) Technology Enhanced Learning. Cham: Springer International Publishing. doi: 10.1007/978-3-319-02600-8.

Holmes, W., Bialik, M. and Fadel, C. (2019) Artificial intelligence in education: promises and implications forteaching and learning.

Phaal, R., Farrukh, C. J. P. and Probert, D. R. (2004) ‘Technology roadmapping—A planning framework for evolution and revolution’, Technological Forecasting and Social Change, 71(1–2), pp. 5–26. doi: 10.1016/S0040-1625(03)00072-6.