Artificial intelligence in the classroom

Ashok Goel, a professor in artificial intelligence at Georgia Tech, has developed a chatbot based on IBM’s Watson to answer the estimated 10,000 questions students ask him and his colleagues each semester. The point is that he didn’t tell his students they were talking to a chatbot, but instead to a teaching assistant called Jill Watson. Nobody discovered the ruse until he told them at the end of term: surprise all round, with some students saying they were going to nominate Jill Watson for best teaching assistant.

I have been teaching for 26 years, and enjoyed almost every minute. I have also spent many hours answering questions online. Aside from my online classes, I have tried to use the online format to supplement my classroom sessions, but I have recently had to stop because I just don’t have enough time.

The questions my students ask are complex, sometimes not that well formulated, and many of them are pretty much the same. Some of them also offer a chance to expand into other areas that might be of interest to them. When you work in what we call participant-centered learning, you soon learn that it’s not about standing up in front of a class and showing how clever you are, but instead your job is to offer students opportunities to learn based on what happens in class, which are very often unique.

When you see that a machine, a chatbot, has managed to take care of a significant chunk of your work, the first temptation is to question its credibility. Don’t bother: Goel’s students didn’t notice, and even if they had, this is still a positive outcome. The important thing here is to remember that if this hadn’t happened by now, it soon would have.

So where do we go from here? Knowing that I can now count on a Watson of my own, and let’s not forget, this machine is the world Jeopardy champion, my approach to teaching could change significantly.

Should I be worried that once my teaching assistant has been replaced by a machine (actually, the school where I teach, IE Business School doesn’t have teaching assistants, and the teacher is fully responsible for both the online and offline courses) that I will be next? I wish that were the problem: I will now start looking for ways to use one. What’s more, if I ever am replaced by a machine, I will have to accept responsibility for failing to understand the changes underway at the moment.

I find the idea of using artificial intelligence in teaching absolutely fascinating, and I’m sure I’m not alone. What better platform could an academic ask for than a machine that can answer questions about his or her field of activity? The question is how universities respond to this change. Perhaps now we have an answer to my complaint: why is it that more than a quarter of a century my classes are still basically the same as when I started: now we machine learning, that complaint takes on a whole new meaning.

My final point addresses this: what will we be able to do when we have interactive knowledge that is able to respond to questions using everyday language? Will teachers become artificial intelligence trainers, helping increase their capacities? Will students interact routinely with increasingly sophisticated chatbots, rather than with flesh and blood teachers? What happens when we think about having intelligent agents that are experts in complex subjects that are able to answer any kind of question? If that’s the future, I can’t wait to be part of it.

(En espa?ol, aquí)