Throughout 2020, people began to associate studying much more with technology, turning to virtual tools that could provide more security with the pandemic in mind. However, the relationship between technology and learning is not only about Zoom. To tell the truth, this seems to be just the tip of the iceberg. And one of the many ways that technology has helped people learn new things is artificial intelligence.
understanding the impact
How Artificial Intelligence has changed language learning involves the level of individual student understanding. AI allows a better student experience by recalling past lessons/lessons, adapting exercises to the needs, and tracking their progress. There are tools like machine learning that create better courses, identify the most challenging material, and prioritize changes.
About what challenges surround this relationship between AI and language learning, AI, by its very nature, sees data as numbers, measures, points, and patterns. That’s not how We think about language. Language is more than a set of words, grammatical rules, and abstract properties: it is a whole communication system connecting with people, expression, and creation.
We can describe AI as a powerful tool to get students to communicate, connect, and express themselves. Even if a machine learning model can tell us the best way to present a new lesson, We are not doing our job as educators if it’s not useful for students. Therefore, in our view, the best AI techniques are those that make our users think, communicate, and use language in meaningful ways.
Language learning in the pandemic
We note that people worldwide are looking for ways to keep busy, especially to feel productive and connected, and that is exactly what learning a language does.
People are using technology to interact with other languages and cultures during the pandemic through music, movies, TV, and online games. Learning a new language through technology has become another way to connect with the rest of the world while being isolated at home.
In March and April 2020, teachers worldwide suddenly had to turn face-to-face classes into fully online classes, which is not easy to do, especially overnight and without resources. Simultaneously, many apps saw the sudden arrival of a new group of users, who may not have been familiar with the content or the technology.
However, with the remote resources, students from all over the world started practicing conversation online.
If you need to know what machine learning means in practice, in everyday life, in learning? The answer is: Machine learning is a way for us to learn about you as you learn. It’s a way to think about data and collaborate with our curriculum experts to improve courses’ effectiveness.
And one of the ways we do that is by monitoring student performance in each exercise. Which sentences are easy for you? Do you have no difficulty with writing, but do you have difficulty with speaking? What kind of grammatical errors do you make? Our algorithm pays attention to all these small details to build a “model,” our best guess about what you know and how well you know it. We then use this model to customize the lessons students see in the app.
In all learning, there is an ideal amount of difficulty: If something is too easy, it gets boring, but if it is too difficult, it is frustrating, and the student gives up. Therefore, there has to be a model to adjust the appropriate type of difficulty for each student to feel the kind of challenge is good, strengthens weaker skills, and does not feel overwhelmed.
Another aspect used by machine learning is A/B testing, used to find out the most effective ways to help people learn and keep going. A/B testing means that before we make any decisions about a new feature, we suggest the change to only a few students first, and then we compare the students in the older version and the students in the newer version. Are there differences between the groups? Does the change make things confusing? Do more students drop out? Does it make people want to take fewer classes or stop studying earlier?
This approach is critical because you can’t see and talk to every student. When learning takes place face-to-face, you can see each student’s face and understand the need to repeat something or overcome frustrations. But for remote learning, especially in an app, motivation and pace are even more critical than in a classroom because now students can just shut down the app if We don’t plan the lesson well. The A/B test is a way for our students to tell us, through their behavior, what helps them learn.
But then the question comes to mind: easier to teach and learn with artificial intelligence? Good. In the scientist’s view, the technology has many advantages, but it also rethinks some teaching parts. One thing that is easy for technology and difficult for traditional teaching is the feedback: if a student is in a class with 20 or 30 other students, they probably won’t get direct feedback from their teacher, simply because there isn’t enough time, but with technology, it’s much easier to provide personalized feedback frequently.
To exemplify, We can be using AI to provide personalized grammar tips directly in a class when a student who is making certain mistakes is noticed. This would be very time-consuming and difficult for a human being, but technology allows us to do this automatically.
The basic chatbot can give multiple-choice activities to a student and correct them automatically, providing feedback. This kind of chatbot-based learning works for language learning and also for other subjects.
But you can also have a more sophisticated chatbot, one that creates a conversation with the student. For example, some startups created a chatbot that allows you to have a dialogue with the student to order a coffee. The chatbot is like a robot that works in a coffee shop. Language learning types of conversational chatbots can be compelling because they allow students to practice conversations without relying on others to speak. The chatbot is a patient practice companion.
However, could chatbots replace teachers? Not quite. We can never imagine a chatbot replacing a teacher because a teacher’s work is about motivation and encouragement; it is emotional and very human work. No AI can replace that. However, chatbots can be great assistants for teachers. They can help by being practice companions for students; they can help automate essential feedback, including speech. And that frees up teachers’ time so they can focus on spending higher quality time with students.
The idea is not to build teacher-robots but teacher-assistant robots. Assistants can help the teacher and the student does simple things in a more automated way so that human time can be spent on more impactful activities.
The main technical challenge is that language students speak with heavy accents and make mistakes, making it very difficult for AI to understand. With data training, you can enhance the AI to get better at this. There are some experiments, but it is quite complex and expensive. And for higher-level students, who need to practice sophisticated sentences and complex grammar, today’s AI is not yet robust enough to give feedback at the same level that a human teacher could. So AI is still no match for a good human tutor. But it can help a lot.