There are few technologies that are as eye-popping as Artificial Intelligence. When you see computers write an essay, recognize human faces, and detect objects, it seems like there are few things a computer cannot do with the help of machine learning (which is a branch of AI). This outward appearance hides a very convoluted mess of math and statistics. It is honestly a web of different implementations, frameworks, libraries, and everything else added together in a soup of technology. The applications of AI that include models that read and summarize texts, or models that can recognize faces look very interesting but the process of creating those models is anything but interesting. Now, a question arises, how can we make learning machine learning as “cool” as its applications so that students do not lose themselves in a web of programming, math, and statistics? Let us investigate how different educational technologies in the classroom try to solve this problem, from educational software that makes learning machine learning easier, to different platforms a teacher can use to upskill their knowledge in AI.

We should start from the basics of what makes modern AI All modern applications of AI stems from a branch called machine learning. Inside this branch of machine learning, there is a sub-branch called deep learning. If you have experienced anything with AI, it is highly possible that it was a product of deep learning. Deep learning utilizes what is known as neural networks to map a set of inputs to an output. What this means is that if you want to train a face-detection model, you will need many photos of yourself in various lighting situations, and poses. The neural network will take all those data and then “learn” your face. The end-product is a model that can recognize your face and nothing else. If you noticed from the example, it is very specialized. A model will never learn something on its own, an engineer will almost always need to feed data into the model.

Learning machine learning requires a foundation of mathematics, statistics, programming, and machine learning-specific libraries. There are too many requirements before a student can start learning machine learning. Machine learning might be a topic that is next to impossible to be taught right with traditional techniques. Educational technologies have devised a plan to circumvent all the steep learning curves and the requirements for learning machine learning. Sites like Machine Learning for Kids, and Teachable Machine are meant for kids to start learning machine learning immediately without the hassle of writing hours of code. They use visual-based programming languages like Scratch or graphical user interfaces to make it easier for students to learn. This way of learning machine learning will give the students an intuitive understanding of machine learning, and a very good foundation.

Modern AI models are not transparent. They are opaque and nobody knows how they work internally, or why the model decided to train itself this specific way and not the other way. The problem now is, how can we teach students something that not even the teacher can fully understand? There have been papers that mention the fact that teachers worry about AI teaching, and have a lack of confidence in their understanding of the topic due to pedagogical information not being readily available in the particular field. This problem can also be solved by educational technology platforms. The popular educational technology platform Coursera provides a course called Artificial Intelligence (AI) Education for Teachers, which in turn is offered by Macquarie University and the corporation IBM. Even teachers can make use of EdTech platforms to upskill themselves in the ever-changing subject that is machine learning.

After an educational institution as the ability to skill their teachers, as well as be able to afford the required IT infrastructure, the next thing is to pump out students who are skilled enough to apply what they learned in the real world. This is one of the hardest things to do in any area of teaching, but it is harder by an order of magnitude when the students do not understand how their own models work from the inside out. The true and tried method to create good machine learning engineers is practice. Google offers everyone free computing powers on their website called Google Colab. Students can utilize that website to practice how to use GPUs to train their own models. This decreases the financial pressure that an institution faces to buy a lot of computers with GPUs. 

Although AI can be a force of good, it can be an even worse force for something bad. Elon Musk’s company OpenAI famously did not release a trained model of their architecture (GPT-2) because it was too good at creating written content that a bad faith actor would be able to create millions of misleading posts. Children from a young age should be taught about the bad effects of unmoderated AI Learning about AI ethics is a very important topic, and there is no better way to teach that than by showing them how AI models affect them online. A website, made by Mozilla, called Track This shows you how well the algorithms on the internet know you. They show this by fooling the algorithm into thinking that the user is someone else. Websites like these, along with others linked in The Algorithm and Data Literacy Program can be used in the classroom to show just how prevalent algorithms are in our lives. It just takes a few engineers in these companies to affect billions of people around the world.

In conclusion, artificial intelligence (which is used interchangeably with the term machine learning) is a technology that is very prevalent in our lives whether you like it or not. The new generation should be taught machine learning so that they are well educated when entering this world. AI is famously a very opaque system where nobody knows how a model works, but there are new technology applications that make the process of learning machine learning for kids easier. When the first step is easy and fun, they are more likely to not only learn more but also be keener in learning how everything functions internally. Another problem that educational institutions face is the fact that their teachers lack the proper resources for themselves to learn about AI This problem is also solved by companies such as Coursera which provides a world-class education for a fraction of the cost. In the end, it is also very important to teach the students about the negative effects of algorithms in the world. There are resources online that show how prevalent algorithms are in our lives and how easy it would be for them to affect billions of lives. Learning the ethics of AI is also a very important pillar in the education of students in the subject of AI.