1. Home
  2. IdeaLab
  3. Artifical intelligence in education: focusing on the greater good

Artificial intelligence in education:
focusing on the greater good


December 15, 2020

The background is navy blue. At the center, a book with white pages is open. White neon icons (from left to right: an apple on a book, a backpack, a pencil, a closed book, and a robot's head) are floating above the open book. Pale blue network links are running behind the icons.

In 2020, one cannot ignore the ever-increasing presence of artificial intelligence (AI) in all areas of society. It is also impossible to turn a blind eye to the societal shifts and the new opportunities AI brings forth in education. AI greatly influences individuals, students, and our society, and all need to develop critical perspectives as to its application in our lives. As UNESCO stated this past March 2020, the systematic integration of AI in education has the potential to address some of the biggest challenges in education today, innovate teaching and learning practices, and ultimately accelerate the progress towards the fourth sustainable development goal (quality education). However, the integration of AI in education is not without its challenges.

Integrating AI into education represents significant challenges for educational systems across all countries:

First, the future generations of scientists must receive adequate training in a context where AI will undoubtedly hold a more prominent place in the world of tomorrow. Instead of dreading the arrival of smart robots that may transform future jobs, it is best to prepare learners for this reality. This would involve making significant changes to educational programs in elementary and secondary schools, especially in the sciences. If learners are trained in AI and its applications, it will help prevent potential technology abuses in the near future.

Second, educational systems are grappling with a new industry of academic resources that rely on AI. But stakeholders are still struggling to make sense of the promising advances of AI (better learning, increased learning, reducing inequities in education, etc.) and the real educational value of these tools.

Third, the issue of teacher training comes with its share of academic challenges. In a context of a severe teacher shortage, how can we think about training them in AI? However, the goal is to initiate them to the basics of AI in education. We can’t ask teachers to become AI experts, but they should at least be prepared to work in schools of the future. At the same time, we must equip our young people with the tools they’ll need to build tomorrow’s society. This aspect is a particularly important challenge for governments; for AI to make a real contribution to the academic success of all students, the role of teachers remains as essential as ever.

Fourth, AI raises ethical issues, which, rightfully, are of concern for many governments. Despite its tremendous potential and growing number of initiatives, AI has made few inroads into the field of education. While many key players in education are interested and curious about AI, they also have quite a few fears and concerns. And education is not free from the fear of AI. Will AI make education more effective, at the expense of the emotional side of teaching? Will robots replace teachers, as suggested by Neil Selwyn in his book Should Robots Replace Teachers1? That fear was reinforced recently in Canada during public consultations on the use of AI for an app that determines the risk level of individuals to be infected with COVID-19. Experts and individuals alike voiced several concerns. Governments and educational institutions must strike the right balance between the industry’s promising advances (better learning, increased learning, reducing inequities in education, etc.) and AI ethical issues of concern for the general population

Fifth, as noted by UNESCO, there is a need for a human-centred approach to AI, one that aims to shift the conversation to include AI’s role in addressing current inequalities regarding access to knowledge, research and the diversity of cultural expressions, and to ensure AI does not widen the technological divides within and between countries. The purpose of this column is to provide an overview of the educational uses of AI. This topic is far too broad to be addressed in a few pages, so we’ve narrowed our focus to help academic stakeholders learn more about AI in education. First, we provide a definition of artificial intelligence. We then address the issue of AI’s impact on education in order to demonstrate the importance of paying particular attention to this issue, followed by the results of a vast worldwide survey that sheds light on the prevalence of AI in education. Many topics are broached, such as the national AI policies of different countries, how AI is being taught to students, and the AI training of future teachers. Finally, a brief section on the guidelines needed to provide a framework of AI in education precedes the conclusion of this text.

What is artificial intelligence?

In a nutshell, it is a machine, system or software that can be taught or learn how to make decisions within specific contexts, without having been previously programmed for the full range of situations. For instance, a so-called smart car can drive down a road it has never taken before. Voice recognition software does not need to hear the voice of a specific person to understand what is being said. Many might wonder whether “intelligence” and “machine” can be uttered in the same sentence. The idea of devices with the ability to think was born in the 20th century, in science-fiction novels and movies. Fictional tales of robot uprisings have been told around the world, which have undoubtedly contributed to fears about artificial intelligence. For instance, there’s HAL 9000, the sentient computer and main antagonist of 2001: A Space Odyssey, and, more recently, Skynet, the artificial intelligence system featured in the Terminator movies.

Nowadays, many believe that reality is creeping closer to fiction. Indeed, smart machines can now best a human at chess, and carry out calculations much faster than any man or woman. And while we can now converse with household devices—smart speakers or virtual assistants such as Alexa, HomePod or Google Home are growing increasingly popular—we are nowhere close to Skynet or HAL 9000. Some tools, software and machines in our daily lives possess some level of intelligence, but it is very context specific. In other words, these tools are not self-aware and cannot learn beyond very specific areas. For example, while a car can learn how to drive, it will never appreciate art. For many, AI is the artificial reproduction of some of human intelligence’s cognitive abilities, for the purpose of creative software, tools, or machines capable of carrying out tasks normally reserved for the human brain. Artificial intelligence therefore translated into computer programs that can learn and put into application the knowledge it has acquired from a phenomenal amount of data, patterns, and models, to solve problems in a given field.

Deep learning is a concept that often comes up when discussing AI. In a way, deep learning mimics the way a human brain functions, as it is programmed to learn in a hierarchical fashion. For example, a cat has a set of typical features: certain shapes for its head, fur, nose, ears, and so on. To recognize an image of a cat, a search engine such as Google interprets a vast number of the elements of these features or inputs. These inputs are weighted for relevance and accuracy as the search engine advances through a series of data layers that grow in complexity and accuracy to make predictions. In the end, without having previously seen all existing images of cats, Google can generally recognize that a cat is a cat. This is a significant advance in AI. There is no longer a need to teach machines everything, as they can now learn and recognize, without having to be trained first. In the case of facial recognition, a feature available on many smart phones, the system analyzes facial traits, facial features, and compares them with any given faces. Many systems resort to deep learning to recognize people, without first having seen the billions of faces that exist today.

AI: what are the benefits for education?

An analysis of the scientific literature sheds light on the many potential benefits of integrating AI in education, depending on the type of use. Six key benefits are outlined below, with examples of tools used in education.

AI and the academic success of learners

For many researchers, the case for integrating AI in education is, first and foremost, to address several of the challenges in the field of education (e.g., school dropouts2, teacher shortages3). It is also meant to provide innovative ways of teaching and learning. A significant advance of AI in education is machine learning. This aspect of artificial intelligence allows a system to automatically generate knowledge by processing collected data. As the system acquires knowledge, it creates models or patterns allowing it to make decisions. For example, the UTIFEN remote learning platform takes all the success and failure pathways of its users into account and then creates “ideal” intervention models for learner success. This “training” prompts the platform to send out automatic and personalized reminders at specific times to learners, therefore increasing their chances for success. Machine learning is a way of learning, without requiring prior programming.

AI for helping teachers with time-consuming tasks

With AI, humanoid robots will play an even larger role in classrooms. Not to replace real-life teachers, as some Hollywood movies may suggest, but rather to help teachers with complex and time-consuming tasks. AI can be of service to teachers by automating time-consuming tasks, thereby freeing up their time to focus on their students.

AI and intelligent tutoring: Immediate and readily available support

AI can create greater interaction between learners and academic content. An example is the chatbot, which is a communication interface between humans and software. Chatbots, similarly to smart speakers like HomePod, Amazon Echo, and Google Home, can recognize the user’s language and simulate a real conversation. AI allows for the implementation of intelligent tutoring platforms for in-classroom and distance learning. A growing number of learning platforms today are resorting to intelligent tutoring. This is a growing trend, and, combined with the rapid expansion of mobile technology, opens up exciting opportunities for learners and educators alike. In France, students can use Jules, a digital homework helper (chatbot) that can answer their homework questions quickly and accurately. In this case, AI is used in the form of a chatbot that can redirect the user to other content. The platform is free and is available on the Web and as a mobile app. Powered by Google AI, the Socratic learning app helps students with their homework by providing them with educational resources. The app uses AI to find the best online resources to help a student learn. Korbit is a personalized learning platform which using AI to help students learn data science skills. This personalized program with one-on-one interactive collaboration based on problem-solving exercises and real-time feedback features personalized diagrams and concept trees to illustrate how concepts are linked together. Korbi, an intelligent tutor, guides the learner throughout the learning process. Offered in six languages, Korbit is available online and as a mobile app.

AI and access to knowledge

Machine learning also benefits automatic visual recognition software. A prime example is the LeafSnap application, which can identify tree and plant species and provide a wealth of information from photos snapped with a smart phone or tablet. As a growing number of schools are now providing tablets to their students, these types of apps carry exceptional cognitive potential, beyond the motivation that this arouses in students. Alloprof, a well-known homework assistance organization in Quebec, launched a new intelligent education platform to help an even larger number of students. Students struggling with math problems can use the Photomath mobile app to snap a picture of the mathematical equation, whether it’s printed or hand-written. Photomath resolves the mathematical equations by providing a step-by-step solution onscreen. This learning platform uses AI to read and resolve mathematical problems and is available in eight languages.

AI and personalized learning

Personalized learning, as we’ve seen with the UTIFEN projects and other tools, is arguably AI’s greatest gift to education4. AI can make learning more meaningful and enjoyable by personalizing learning exercises5. The language learning system Duolingo is an outstanding example of personalized language learning. With its voice recognition system and more than 600 million users worldwide, this AI-powered tool is widely used in the field of education. The intelligent language learning platform is one of the most popular in the world and has found its way into classrooms, where it is used by thousands of teachers to improve their language lessons. The app’s AI can track students’ existing knowledge of a language and provide personalized learning activities. Other language learning platforms such as AI Grammar Checker offer personalized help with spelling adapted to the learner's existing level of knowledge. With the Mathia platform, teachers can build individual paths of learning for each individual student based on their strengths and weaknesses with the help of a 3D visualization tool. An AI system provides an overview on the progress and acquired knowledge of each individual student. While working on mathematical problems, students have the support of an endearing character called Mathia. The platform also provides pupils with a 3D visualization of mathematical concepts. Mathia is a free Web-based application for teachers, available in English and French.

AI and learning assessment

AI allows for the automatic correction of certain kinds of schoolwork, which frees up teachers’ time for other academic tasks. However, existing thesis correction apps leave something to be desired. Despite the amazing progress made, human intervention remains essential6. AI facilitates the continuing academic assessment of learners. Thanks to AI, learners’ experiences can be tracked throughout their entire learning journey, and their level of competence can be assessed with relative accuracy at any given time7. Feedback is an integral part of learning assessment. AI increases opportunities to provide educational feedback to students. For example, the UTIFEN platform sends personalized texts to students throughout their learning journey. With AI, not only is the feedback personalized, it’s faster and more frequent. The Turnitin software uses AI to detect degrees of plagiarism in students’ work when they “turn it in”. It shows the parts of the student’s work that are likely to have been plagiarized, the potential sources, and the percentages of these sources that have been plagiarized.

AI in education: glance around the world

National AI policies

A large-scale international survey was held recently to better understand international initiatives in terms of AI and education. The OECD AI Policy Observatory is a valuable resource for identifying such initiatives, even if many of the initiatives surveyed do not focus on AI in education per se. More than 16 countries have implemented national AI strategies with a focus on education. In 2017, the Government of Canada mandated the Canadian Institute for Advanced Research (CIFAR) to develop and lead a Pan-Canadian Artificial Intelligence Strategy. One of the objectives of CIFAR is to increase the number of outstanding artificial intelligence researchers and skilled graduates in Canada. Indeed, training and fostering AI talent is now considered an urgent matter in many other G20 countries. Such initiatives are found in Germany, Saudi Arabia, Australia, Belgium, China, South Korea, several U.S. states, France, India, Italy, Japan, Mexico, the United Kingdom, Russia, and Switzerland. In many of these countries, the initiatives take the form of specialized university courses, such is the case in France and in Canada.

Are students learning about AI?

As AI continues to pervade a growing number of areas in our daily lives, it appears necessary that it become a focal point in education. Schools have no choice but to actively participate in training citizens of tomorrow and key stakeholders in this technological advance. Similarly to training in computer programming, which has seen a boom in recent years, the goal is not to train students to become computer engineers, but rather to ensure they are fully prepared for the world of tomorrow. Their command of this digital component must also allow them to deal with technological advances that will arise in years to come, especially when it comes to inroads in artificial intelligence. By learning about AI, they will also develop critical perspectives on these advances. A recent survey allowed us to see the extent to which AI is being taught to students at the elementary and high school level. Seven states have made AI learning a mandatory component of their curricula: Australia, Canada, China, the U.S., India, Wales, and the United Kingdom. Seven of the 10 Canadian provinces have integrated AI into their educational programs. Therefore, (1) in New Brunswick, students must be able to identify the key elements of AI; (2) in Prince Edward Island, learners discuss the ways AI will affect science and engineering; (3) Saskatchewan students analyze AI uses in robotics; (4) in Ontario, learners must be able to assess and discuss emerging technologies, including AI; (5) in Newfoundland and Labrador, when pondering the prevalence of computers, students are called upon to discuss AI-related issues; (6) in British Columbia, AI is one of the current and future technologies to consider when developing concepts and content; and finally, (7) in Quebec, the Digital Competency Framework, the latest governmental policy on the uses of technology in education, states that it is important to develop, as early as the elementary school level: “a general understanding of artificial intelligence and its impact on education, society, culture and politics.” (p. 14). In Australia, students are given the opportunity to analyze a range of digital solutions, including AI systems. Their curriculum even states that students must attempt to resolve complex problems using AI. In China, schoolchildren are initiated to the basics of AI as of the age of 5. In the United States, 32 states have integrated AI into their educational programs. In India and in Wales, artificial intelligence is just one of the digital components included in those countries’ curricula. And in the United Kingdom, secondary school learners must be able to define AI as well as the key features of neural network models.

Are teachers being trained in AI?

Not only is AI a powerful tool that must be integrated in educational systems, but educational institutions must also actively participate in preparing future teachers for AI. Why should teachers-in-training be prepared to work with AI in education? Because AI carries a great influence on individuals and societies, and we need to develop critical perspectives on AI issues. Because if teachers are trained in AI, it may help prevent potential technology abuses in the near future. Because for AI to make a real contribution to the academic success of all students, the teacher’s role remains as essential as ever. Because intelligent robots will transform the jobs of tomorrow, children should begin preparing for the new reality as early as elementary school. Because tech players should not be the only ones to have a say in the matter. Recent studies show that a dozen countries are or are planning on integrating artificial intelligence in teacher training programs: Germany, Australia, Belgium, Canada, China, South Korea, the U.S.8, India, Italy, the United Kingdom, and Russia.

A framework for AI?

Many experts, such as Yoshua Bengio of the Université de Montréal9, believe that AI may become dangerous if society fails to impose limits on intelligent machines and their applications. Many initiatives have been developed to ensure a proper and strict framework of AI uses. One of the first was the Montreal Declaration for the Responsible Development of Artificial Intelligence, which has three main objectives: (1) Develop an ethical framework for the development and deployment of AI; (2) Guide the digital transition so everyone benefits from this technological revolution; and (3) Open a national and international forum for discussion to collectively achieve equitable, inclusive, and ecologically sustainable AI development.

UNESCO recently launched a global online consultation on the ethics of artificial intelligence, to give everyone around the world the opportunity to participate in the work of its international group of experts on AI. This group has been tasked with producing the first draft of a recommendation on the ethics of AI, to be submitted to UNESCO Member States for adoption in 2021. If adopted, it will be the first global normative instrument to address the developments and applications of AI. A first version of the draft recommendation on the ethics of artificial intelligence is available online. It includes many of UNESCO’s concerns, stating that while AI systems can be of great service to humanity, they also raise fundamental ethical concerns, for instance regarding the biases they can embed and exacerbate, potentially resulting in inequality, exclusion and a threat to cultural and social diversity and gender equality; the need for transparency and understandability of the workings of algorithms and the data with which they have been trained; and their potential impact on privacy, freedom of speech, social, economic and political processes, and the environment. These initiatives, especially this latest UNESCO recommendation, seeks to ensure that artificial intelligence is used in education for the greater good.

AI in education: focusing on the greater good

Rather than see AI in education as a panacea or the Holy Grail, we should look at it as a tool with unbounded potential in education. A key challenge for education systems today is to find the right balance between time-honoured teaching practices that have been handed down through the ages and the exciting new opportunities provided by AI. Not just an innovative educational tool, AI, when combined with thought and imagination, has the power to completely transform the learning experience. Moving beyond the school context, AI can help students understand and respect themselves as human beings and fellow citizens of the world. The integration of artificial intelligence in education has the potential to address some of the biggest challenges in education today. A human-centred and well-defined approach to AI should adhere to the mission of education—instruction, socialization, qualification—with the goal of fighting knowledge access inequality and improving human capacity, all while protecting human rights within the context of effective human/machine collaboration in learning, in line with SDG #4.


References

1. Selwyn, Neil (2019). Should Robots Replace Teachers? AI and the Future of Education. Indianapolis: John Wiley & Sons.

2. Sara, N.-B., Halland, R., Igel, C., & Alstrup, S. (2015). High-school dropout prediction using machine learning: A Danish large-scale study. In ESANN 2015 proceedings, European Symposium on Artificial Neural Networks, Computational Intelligence (pp. 319-24).

3. Edwards, B. I., & Cheok, A. D. (2018). Why not robot teachers: artificial intelligence for addressing teacher shortage. Applied Artificial Intelligence, 32(4), 345-360.

4. Yu, D., Ding, M., Li, W., Wang, L., & Liang, B. (2019, July). Designing an Artificial Intelligence Platform to Assist Undergraduate in Art and Design to Develop a Personal Learning Plans. In International Conference on Human-Computer Interaction (pp. 528-538). Springer, Cham.

5. Javaid, Q., Arif, M., Talpur, S., Korai, U. A., & Shah, M. A. (2017). An intelligent service-based layered architecture for eLearning and eAssessment. Mehran University Research Journal of Engineering & Technology, 36(1), 97.

6. Luckin, R., & Cukurova, M. (2019). Designing educational technologies in the age of AI: A learning sciences‐driven approach. British Journal of Educational Technology, 50(6), 2824-2838.

7. Luckin, R., & Cukurova, M. (2019). Designing educational technologies in the age of AI: A learning sciences‐driven approach. British Journal of Educational Technology, 50(6), 2824-2838.

8. At the time of the census, in five U.S. states: Alabama, Arkansas Florida, Indiana, and West Virginia.

9. Bilodeau, Maxime (2018). Monsieur intelligence artificielle, Québec science, January-February, p. 61.


Author

Thierry Karsenti, Ph.D.

Canada Research Chair in Technologies in Education, Université de Montréal, Canada