What Will AI Do for Students and Teachers?

Special Focus : Reinventing Education for the AI Age
September 04, 2017

The hype is never far away from Artificial Intelligence. Back in 1953, Marvin Minsky, a founding father of AI, declared: “We’re going to make machines intelligent. We are going to make them conscious,” to which his contemporary Douglas Engelbart replied: “you’re going to do all that for the machines? What are you going to do for the people?”

 
Sixty years on, Engelbart’s question remains pertinent to EdTech innovators. Artificial Intelligence offers unprecedented opportunities to deliver learning experiences tailored to the individual needs of every learner. But those opportunities bring unprecedented risks, and while the apocalyptic warnings around AI may lie beyond Education’s remit, blind deference to these technologies only perpetuates flawed pedagogies. The impact of AI on Education will depend on how we understand the relationship of students and teachers with machines.

Personalised learning is widely heralded as education’s liberation from the tired convention of one-size-fits-all. Traditional models of schooling cannot cope with the growing demands of students and teachers. The spread of knowledge and skills among learners, along with rising class sizes and relentless workload, make it almost impossible for teachers to devote the individual attention every student requires.
 
Virtual tutoring is the great enabler of personalised learning. For the first time in history, we have the tools to extend the benefits of one-to-one tutoring to all learners. A virtual tutor is a fully automated instructor that uses AI to simulate the behaviour of a human tutor: it starts by diagnosing each learner’s specific knowledge gaps across different curriculum areas and then develops individualised learning pathways to fill those gaps, constantly adapting the difficulty and form of content based on learner responses. 

The learning gains are impressive: the Maths-Whizz virtual tutor developed by Whizz Education has been shown to advance students’ mathematical knowledge by 18 months in their first year with just an hour of use each week.
 
It can make for nervy reading if you are a teacher but rest assured, the role of human educators is more vital than ever. Virtual tutors are limited in two key areas. First, they are governed by data-driven measurements that capture only a sliver of students’ thinking. Virtual tutors may impart core knowledge with frightening accuracy, but they do not broach deeper aspects of cognition like problem solving or reasoning; the very skills that are gaining currency as AI sinks its claws into wider streams of blue-collar work. Even more importantly, virtual tutors cannot provide the social or emotional connection that underpins the richest learning experiences. No machine learning algorithm can replicate the human dimension of teaching – the part that makes personalised learning ‘personal’.

Unfiltered acceptance of technology will result in the same narrow and impersonal learning experiences that arise from one-size-fits all instruction. The key question then, is how teachers can marshal virtual tutors to deepen their interactions with students.
 
Artificial intelligence is a loaded term because it connotes substitution of human labour. Engelbart himself preferred the notion of augmented intelligence: using technology to advance our distinctly human capabilities. In education, augmented intelligence recognises that while technology can accelerate better learning outcomes, good pedagogy will always be the driver.
 
Too much of teachers’ time is wasted on mundane tasks that distract from their face-to-face interaction with students. Virtual tutors, and the real-time learning analytics they supply, automate menial components of teaching such as knowledge transmission and grading of structured tasks. Teachers can instead focus on planning and deliver richer classroom experiences, armed with data-driven insights. The real power of virtual tutoring is in augmenting the teacher’s role by allowing them to devote more time to each learner.

Data can only capture so much. Education is already in the grip of high-stakes measurement as students, teachers and entire education systems are increasingly judged solely in terms of blunt performance outcomes. AI has the potential to disrupt the assessment paradigm: instead of relying on snapshot test scores, educators can track students’ progress as it unfolds, thereby lowering the stakes by using assessment data as a proactive feedback tool rather than as an accountability mechanism. But these tools may also be exploited to advance the high-stakes testing agenda. Big Data represents big opportunities to those who benefit from the commoditisation of learning and teaching. Data only ever represents a partial measure of students’ learning; AI must not be allowed to substitute for the human judgement of teachers.
 
The affordances and scalability of virtual tutors means their place in Education’s future is inevitable. But the manner in which these technologies are designed and implemented is a conscious choice for educators. AI can amplify broken educational practices just as well as it can accelerate the noble goals of personalised learning. Unbridled enthusiasm for AI must make way for a critical reflection of what these technologies can do for the students and teachers they supposedly serve.