hide announcement WIN your very own meteorite! Subscribe, gift or renew a subscription to Cosmos and automatically go into the draw – Shop now!

Teacher robots are coming but they are not who you think


Replicating a good teacher’s skills is a challenge for artificial intelligence, but as classroom back-up team, ‘cobots’ are hard to beat, say Leon Sterling and Mike Timms.


Nao in classroom – BSIP / UIG / Getty images

AlphaGo, Google’s AI software, last week beat the European champion at the ancient board game Go, a complicated strategic game long regarded as a major challenge for Artificial Intelligence researchers.

The breakthrough came far sooner than AI experts had predicted.

If previous AI triumphs are anything to go by, our awe over this feat will quickly segue into uneasiness, as professionals worry about artificially intelligent robots taking over their jobs, the very jobs once thought immune to the advances of human-built machines.

With the proliferation of online courses, and other IT pedagogical tools, teaching is fertile ground for these concerns to flourish.

We think teachers can relax. Sure, artificially intelligent robot teachers are coming to a classroom near you. But they won’t be making teachers obsolete anytime soon.

Good teachers are the quintessential all-rounders. They must know their subject, be skilled communicators, be able to interpret the nuances of different children’s behaviour, and show empathy. Teachers need a highly developed theory of mind (the ability to put yourself in another person’s headspace) so that they can engage with a student as they learn new concepts, even gain entirely novel insights.

By contrast, AlphaGo is essentially one-dimensional. It has a deep understanding of patterns in Go games but little else.

To succeed in the classroom, an intelligent robot would need to master the even more complicated task of natural language processing at the level required to speak with a class of children.

So far, none has got close.

Intelligent machines also struggle at recognising symbolic representations such as handwriting, mathematical notation, or a nine-year-old’s scrappy diagrammatic explanation of photosynthesis.

Ditto detecting the emotional state of each one of a group of learners — critical to effective teaching.

When it comes to holding a students’ attention, robot interfaces — even ones that use fairly rudimentary dialogue and gesture — can be an advantage.

Nevertheless, AI and robotics are on the verge of transforming the classroom — just not as a single all-purpose replacement robot teacher, but as “educational cobots” – robots designed to work with humans, rather than instead of them.

Take tailored learning.

Knewton, a company founded in 2008, already offers adaptive learning software that diagnoses learners’ needs and changes the sequence of lessons for individual students in mathematics, science and other subjects.

At the moment, Knewton, and other similar applications, interface with the learner through the screen and keyboard.

But when it comes to holding a students’ attention, robot interfaces — even ones that use fairly rudimentary dialogue and gesture — can be an advantage, as one of us has shown with a humanoid robot called Nao.

Holding students’ attention is a precursor to learning, including re-learning physical skills following severe injury.

In research at Melbourne’s Royal Childrens Hospital, Nao helped children recover from injuries by encouraging them to do exercises and relieving the monotony and struggle of the effort.

The children found the robot engaging and amusing, and they were motivated to copy and outperform the robot in friendly competition.

Repetitive learning tasks are also core to certain stages of learning in the classroom.

But the tasks can challenge the attention spans of both learners and their teachers — engaging, motivating robots like Nao to the rescue.

Another challenge teachers face in classrooms of 25 or more students is how to monitor all that is happening, especially when students are working in small groups.

Having a roving cobot in the classroom observing the students and notifying the teacher when extra help is needed would be an advantage.

This extra support is on the horizon as robots’ ability to recognize faces, and software to detect basic emotions from facial expression, steadily improves.

The repertory of cobots in the classroom will no doubt expand, as advances in AI and robotics continue to expand our imaginations.

Related reports:
Learning to live with robots
ArtificiaI intelligence: Your future today

Leon Sterling is a computer science and researcher.
Mike Timms is Director, Assessment and Psychometric Research at the Australian Council of Educational Research.
Latest Stories
MoreMore Articles