Whether you’re fearing or dreaming of the vision of AI robot tutors, it’s time to hit pause and come back to reality.
Artificial intelligence will very likely shake up education, but not in the way or at the speed that many of its proponents or detractors imagine. While technology continues to hold promise for enhancing education, there’s a growing risk that we will lose sight of the fundamentally human nature of how we learn.
For my recent report, Exploring Edtech and AI in Maths Teaching, I spoke to hundreds of people: teachers, students, policymakers, tech developers and many more. The resounding message is that the edtech hype party is still in full swing, but doubts are creeping in around the edges.
The problem with AI in education
Many proponents of AI like to bang the drum that we should create tech-powered tutors, citing a historical study from the 1980s by Benjamin Bloom (of taxonomy fame).
But the paper itself does not stand the test of time, with newer studies (such as those from Matthew Kraft and his colleagues) showing the fiendish difficulty of scaling effective tutoring, even tutoring led by humans.
When tech-based tutors were tried at Khan Academy, enthusiasts noted happily that the students who used the platform properly showed significant progress (an effect size of 0.26). But critics rapidly pointed out that the researchers had dismissed 95 per cent of students for not using the system properly.
More on AI in education:
The trouble is, a lot of technology requires students to interact mainly with a screen and to be able to maintain their own focus, confidence and pace. But this neglects so many aspects of real human learning.
For example, online one-to-one learning systems generally don’t allow time for exploring multiple views, creating high-quality dialogue, debate or discovery, unlike a classroom with a skilled teacher.
And it’s sometimes the discussion that is important, not just consuming content: anyone who has clicked as fast as possible through an online mandatory training course will rapidly spot the difference between merely being exposed to content and having engaging experiences discussing and exploring it.
Attention and retention
Learning on a screen can also be a poor relation to traditional methods: some studies suggest that reading and writing on devices can lead to lower attention and retention than reading and writing on paper.
But the techno-pessimists also need to pause for thought. Some platforms are offering significant support in setting and marking homework, giving students more tailored support and giving teachers rapid, workload-free insights to use in lesson planning.
Some subject areas are using technology not just as a means of learning but as an outcome of it, educating students to carry out key processes with some of the latest software, coding or data manipulation tools.
What seems clear to me is that human relationships, whether teacher-to-student or among student peers, will continue to play an irreplaceable role in effective education. We therefore need to focus more on finding ways to deeply integrate tools into effective classroom practice, rather than trying to replace teachers and peer interactions.
To do that, we need to consider the barriers and the opportunity costs. Imposing technology where it isn’t needed just adds costs of purchase, costs of upkeep, stress and distraction.
There are also very, very few technologies that are so wildly impactful that it’s definitely worth disrupting the expert routines of some of our most effective teachers, so we need to be cautious about blanket imposition, given the precarious nature of teacher morale and retention.
Finally, we need to start getting a grip on the lack of quality digital infrastructure in schools, including ensuring that all school leaders have the training and support they need to commission, deploy and maintain the complex IT systems that every school now has.
David Monis-Weston is AI lead at Purposeful Ventures and a former maths and physics teacher