AI in Education: Theory and Practice

Experience AI: Experience AI is an educational programme that offers resources on artificial intelligence and machine learning for teachers and students aged 11–14. Developed in collaboration by the Raspberry Pi Foundation and Google DeepMind, the programme supports teachers in the exciting and fast-moving area of AI, and gets young people passionate about the subject. Includes free lesson sets complete with lesson plans, presentations, videos, and worksheets. (Registration required)

Day of AI curriculum: Developed by leading faculty and educators from MIT RAISE (Responsible AI for Social Empowerment and Education), each curriculum features a series of 30-60 minute lessons that engage kids in creative discovery, discussion, and critical thinking as they learn the fundamentals of AI, investigate its societal impacts, and bring grade-relevant applications of artificial intelligence to life through hands-on activities that are accessible to all, even for those with no computer science or technical background. (Registration required)

Code.org AI curricula: Includes short "How AI works" videos and free lessons/teaching units for all age levels on a range of topics from "AI for Oceans" to "AI Ethics"

Machine Learning for Kids UK:  A free tool introducing machine learning by providing hands-on experiences for training machine learning systems and building things with them. It provides an easy-to-use guided environment for training machine learning models to recognise text, numbers, images, or sounds.

I Am AI: Resources to read, build, make music, experiment and play with AI.

AI can be used to develop accessible technologies such as voice recognition systems and automatic translation APPs or devices, enabling students with different needs and levels of competence to interact effectively with educational tools and resources and empowering them to actively participate in learning.  AI-assisted learning is another strand through which inclusion is promoted;  AI-based virtual assistants can offer students permanent personalised support during the creation and problem-solving process, providing step-by-step guidance, useful tips and immediate feedback suited to their level of competence and learning style.

In addition to other advantages such as supporting more inclusive assessment processes, we might also mention that the introduction of AI in makerspaces and classrooms even contributes to reducing the gender gap in STEAM areas. In February 2016, at a TED talk, Reshma Saujani, founder of the “Girls Who Code” programme, said that the large deficit of women in STEAM areas was due to the fact that we are educating boys to be brave and girls to be perfect, reaffirming the urgency of changing the paradigm.^[1] With the introduction of coding and AI, this gap can be narrowed, as the methodology requires perseverance and shows that mistakes are not “fatal”; on the contrary, it demonstrates that success is a path of corrected mistakes, helping to demystify the idea that girls have to be perfect.
 

“Ethics guidelines for trustworthy AI”, publication of the Directorate-General for Communication Netzworks, Content and Technology of the European Commission, 2019, and “Ethical guidelines on the use of artificial intelligence and data in teaching and learning for educators”, publication of the Directorate-General for Education, Youth, Sport and Culture of the European Commission, 2022.

"Artificial intelligence in education: challenges and opportunities for sustainable development", UNESCO Working Paper on Education Policies, 2019, full text available online.

"Ethical AI for Teaching and Learning", Center for Teaching Innovation, Cornell University.

"The Green Dilemma: Can AI Fulfil Its Potential Without Harming the Environment?", Earth.org, July 2023.

A Pidgeon's Tale. A Comic Essay on Artificial Intelligence and Sustainability, by Dr. Julia Schneider, illustrated by Pauline Cremer (retrieved on July 16, 2024).

In this approach, perception refers to the ability of AI systems to recognise and understand the world around them in the same way as humans, which involves processing sensory data such as images, audio and/or text to extract meaningful information. For example, in computer vision, AI algorithms can be trained to recognise objects, faces or patterns in images, using techniques such as convolutional neural networks. In natural language processing, AI systems can analyse and understand the meaning of written or spoken text, enabling automatic translation or the extraction of information from large volumes of text. This ability to perceive is fundamental to many AI applications, from voice recognition systems in virtual assistants to computer vision systems in autonomous vehicles.

The second idea proposed refers to representation and reasoning, which are essential for AI systems to be able to organise knowledge in a meaningful way and make autonomous  decisions based on that knowledge. In symbolic artificial intelligence, systems represent knowledge using symbols and logical rules, enabling deductive reasoning. For example, in expert systems, knowledge is represented as a set of production rules, which are used to infer new information from input data. In deep learning, on the other hand, systems learn high-level representations of data through layers of processing, enabling inductive reasoning in tasks such as pattern recognition or prediction. This idea is central to the development of AI systems capable of solving complex problems and making autonomous decisions.

Learning is one of the most powerful capabilities in AI, allowing systems to improve their performance in specific tasks based on experience. There are different types of learning in AI, including supervised, unsupervised and reinforcement learning. In supervised learning, systems are trained on labelled input-output pairs, allowing them to learn to associate inputs with desired outputs. For example, in image recognition, a system can be trained on a set of labelled images to recognise specific objects. In unsupervised learning, systems learn patterns and structures in the data without external guidance, allowing for the discovery of insights and groupings. In reinforcement learning, systems learn to make decisions and take actions in an environment to maximise a reward, using techniques such as Q-learning or genetic algorithms. This ability to learn from experience is fundamental to the development of AI systems that can adapt and improve over time.

Natural interaction, the fourth idea proposed, presupposes security and ethics as critical considerations in the development and use of artificial intelligence, given its growing influence on various aspects of society. Security issues include protection against cyber attacks and guaranteeing the integrity and reliability of AI systems. Ethical issues involve concerns about algorithmic bias, data privacy, fairness and transparency. AI algorithms can reflect and even amplify existing biases in training data, leading to discriminatory decisions in areas such as hiring, business transactions or criminal justice. What's more, the use of personal data in AI systems raises concerns about privacy and the protection of individual rights. It is crucial to address these safety and ethical issues to ensure that AI is developed and used in a responsible and inclusive manner.

Finally, the social and economic impact of AI is broad and far-reaching, affecting areas such as the labour market, education, health or politics. AI-driven automation is transforming the nature of work, with potential effects on employment, wages and income distribution.