Students Teach Bots - And Learn

Yael Cohen tells how educational bots are utilizing personalized learning with the Protégé Effect and Learning by Teaching

“The classic direction in which learning chatbots or avatars are used is as teachers who give private lessons, as a substitute for a teacher, or as help for a teacher or a TA. But you can use chatbots as a tool that you need to teach, instead of learning for yourself,” says Yael Cohen, an information scientist with a master’s degree in education and technology, who deals with artificial intelligence and personalized learning. “Educational chatbots are an excellent way to take advantage of the Protégé Effect, an educational strategy that in a nutshell claims students learn best when they teach others. Chatbots are a scalable way to achieve this very convincingly – they are used as ‘teachable agents’, a path they has rarely been taken so far. When I studied at Stanford University until recently, I have seen several examples of this, or at least some such experiments, specifically dealing with mathematical instruction for K-12 students.

“There is a doctrine in education called Learning by Teaching. They took very weak children from low socioeconomic classes and had them teach the teacher in a kind of a role-playing game. If they buy into it, they feel a different kind of responsibility, which greatly enhances their achievements. The concept is that there is also a world of accountability – the child’s ability to feel responsible for learning, also from a place of motivation. It was linked to a framework in which you are allowed to fail. That’s true for a chatbot – a safe environment for failure.”

Get your copy >>> Chatbot Magazine

A chatbot that doesn’t let you make mistakes can’t function.

“True, technologically this is what happens. This is important, certainly with the whole trend of Growth Mindset in education, which says that what’s important is that the child and others around him believe that he is capable of growing. Today the most influential factor in learning is your sense of ability. Many very successful children get stuck because they feel failure equals stupidity. By teaching someone, it gives you responsibility, accountability, raises motivation and allows you to err on yout way to understanding how to teach.  In general, it has proven to be much more effective learning. And what you do with it with the using the chatbots – you need to make a chatbot that can make the child teach it. Another way is to create an avatar for you – instead of a private tutor who teaches you, you go to the world of collaborative learning, you’re in a group and the bot is someone you learn with. The avatar tells the kid, ‘Oh, I don’t get it, can you explain it to me?’ In this world, the child has the feeling that he is teaching the avatar.”

The chatbot pretends to be a fool or presents himself as ignorant, while giving all the answers and directing the child their way?

“The computer doesn’t really need to know the answers to the questions, it’s a very different approach from a chatbot, which is a private tutor, a TA, who needs to direct the child. Here it doesn’t need to know all  the answers –  it pretends to be another child, responding as another child who doesn’t understand. That’s what the Protégé Effect is about.”

But where does the child bring the knowledge from?

“The whole thing is within practicing. The child has to understand things for himself. He has the textbook, but instead of going through all the questions and answering them, this is a completely different world – he needs to reach a higher order thinking, a deeper learning, where you need to know enough to teach, instead of saying, ‘It seems to me that this is it,’ when someone else asks him ‘why did you choose this – prove you’re right’ – so he has to say that he teaches how he did it. This is the more innovative usage of chatbots I’ve seen. I do not know to what extent it exists in the market, but pedagogically I strongly believe it.”

Get your copy >>> Chatbot Magazine