The automation debate

What can Artificial Intelligence bring to learning? Edtech professionals Rosamund McNeil and Henry Warren share their thoughts

Rosamund McNeil, Head of Education, NUT, argues that, although technology can be used to enhance the learning experience, the automation of teaching is not in pupils’ best interests…

Clearly there are aspects of education which can absolutely be enhanced by technology. Virtual Learning Environments (VLEs) have become very useful tools for teachers to share resources and information with students, parents and each other. Programmes such as ‘mymaths’ have become increasingly popular in schools, allowing students to revise topics and practise questions, and teachers to set and monitor homework and progress. Tools such as these are used by teachers as an enhancement for learning. Technology alone cannot effectively replace all aspects of a teacher’s role.

Some suggest that the automation of tasks which do not need emotional intelligence could reduce teacher workload. In practice, it is difficult to see where this could happen effectively. Marking often has the largest impact of any such tasks on teachers’ time and there are examples of automated marking systems in use within some schools. However, these systems can usually only say if an answer is wrong and not why.

We are unlikely to ever have technology that can read sentences and determine contextually if a student has written an answer which demonstrates understanding. Even if we were to find a way for artificial intelligence to suitably mark prose, a major issue would arise from this: without reading and marking the text themselves, the teacher would miss vital information about the progress made by the student or class in question.

In terms of the frankly more dangerous concept of automating the act of teaching, there are many points surrounding this which it does not take an educationalist to appreciate. Effective learning occurs when strong relationships are built between teachers and students and the same can be said for creating positive, collaborative working environments. Classroom practitioners are consistently encouraged to display enthusiasm and model what good learning looks like. How could any of these things be effectively replicated by a machine?

Many within education fear that the teaching profession is not held in as high a regard as it deserves. The mere suggestion that you could replace a highly skilled and professionally trained human being with some kind of automated dissemination of information, whilst still aiming for and achieving positive outcomes for young people, might indicate that these suspicions are correct.

Automating the act of teaching could risk reversing the work done in recent years to support the wide range of pupils’ interests and to ensure that the specific needs of pupils with SEND or EAL are supported. How would an automated teaching model facilitate the needs of these students or the additional nurturing and personalisation required by those who lack an inherent motivation for learning?

Even if we were to ignore the importance of forming positive relationships, learning environments, the uniqueness of every young person and the motivation of students, and accept for the sake of argument that teaching could be automated suitably enough to carry out a simple transfer of knowledge, would we actually want this as a society?

Surely we want a world and an education system where our children are treated as individuals with all the emotional and social benefits and developments which occur within classrooms with fully qualified teachers and support staff. 

Henry Warren, consultant, edtech investor, and former Director of Innovation at Pearson & GEMS explores how AI could improve education globally, and how teaching could be automated…

1.6 million…that’s how many teachers we’re missing, globally, in 2017. In all but a few countries education departments’ budgets are stretched. The UK is itself under huge financial stress, with 18% of schools reported to be in deficit in 2017. 

Put yourself in the education minister of Tanzania’s shoes. She has twice the number of children as the UK but only 2% of the budget. It’s worth noting, that according to the latest projections, by 2030, the majority of learners on Earth will be in Africa. 

Too often when discussing AI in education we start in the wrong place. 

We imagine humanoid robots – a ‘Robo Mr Jenkins’ – teaching algebra in Guilford. AI and automation needs to be set in a global context. There is a desperate, unmet need for quality education across the developing world, and change will start here.

But, for a second, let’s put aside whether robots in education are desirable and see if it’s actually feasible.

Ask a friend, in any industry, if a robot could ever do their job. They will tell you (very firmly) – no. Their job is too complex, too varied. There are parts that can never be done by a robot, however advanced the AI. Here’s the thing, they’re both right… and wrong.

You see robots don’t ‘do’ jobs. They ‘do’ tasks. Ask that friend again. Ask them if there are any tasks in their job that could be potentially automated? Most will say yes. And that’s how automation is taking over industries. 

The manufacturing jobs that Donald Trump bemoans have gone to China, in the most part, haven’t. They were simply broken up into different tasks. Tasks that were easy to automate, have been given to machines. Tasks that were harder, were reconstituted into new human jobs, just fewer of them and they were only available to highly skilled people.

Is teaching the exception that proves the rule? Teaching is very complex; many skills intertwined, from empathy and non-verbal communication, through to subject knowledge and more. 

It’s certainly not the easiest set of tasks to automate. However, if we split these tasks into a series of smaller ones, even in 2017, AI can do almost every one of them. In some cases, better than a human. Can we stitch them together to create ‘Robo Mr Jenkins’ that can do all these tasks exactly as a human would? No, we’re at least a decade off that. Probably more. But that may actually be less important than one might think.

So what does this mean for our teaching? We have the exciting ability to redefine what it means to be a teacher in a way that works for both student and teacher. For starters, I see no reason why the teacher of the future does any marking or any admin. But then come the more difficult decisions. Does it make sense that, at any given moment, we have a million teachers all teaching isosceles triangles? Would it be better for us to curate the best lessons and just work with those? What might we lose by doing that? What might we gain? Might teachers move much more towards being learning guides and mentors rather than custodians of knowledge? Does that empower or detract? Does that create more jobs or less? All this is up for grabs, all yet to be defined. 

One thing we do know is that teaching will change – it’s inevitable. It’s now up to us to ensure we preserve what’s great about teaching, yet not be afraid to rethink key aspects. We need to move fast though. Over a billion children depend on us making the right choices.