Why AI should force educators to rethink SDG4

  • January 21, 2020

The new year has brought with it a renewed resolve to address education’s most pressing challenges. The UN has declared this the Decade of Action. With ten years left to achieve the Sustainable Development Goals, what lessons might we draw from the decade just gone? As the world hurtles towards unchartered frontiers, spurred on by the frantic pace of technological development, what challenges should we anticipate, and what are the implications for educators?

To put it mildly, the twenty-teens was something of a mixed bag for education. EdTech writer Audrey Watters has written at length of the failure of EdTech to deliver on its lofty promises. Her list of “100 worst Ed-Tech debacles” is a sober reminder of how technology-led approaches to education are invariably doomed to fail students and teachers.

It hasn’t been all bad: this was the decade, after all, in which the Sustainable Development Goals came to be, as an upgrade of the Millenium Development Goals. For educators, SDG4 marked an intent to focus not just on educational access, but also quality. We are a long way off that ideal, with an estimated 617 million students lacking proficiency in the core literacies of reading and numeracy.

We have to reckon with two realities: 1) without wide-scale deployment of technology, the SDGs will remain an unattainable ideal and 2) technology poses as many threats as opportunities, never more so than in matters of learning and teaching. How, then, do we bring technology to bear on SDG4 without perpetuating the mistakes of past implementations?

 

 

AI and SDG4: A help and a hindrance

At the heart of this question is the role of Artificial Intelligence (AI), a highly contested term and a field with its own chequered history. What is beyond dispute, by now, is the pervasive influence of these technologies in all corners of society. It’s not a question of whether to adopt AI, but how to do so in a manner that advances rather than undermines our educational objectives.

A recent paper published by Nature that examines the role of AI in achieving the SDGs concludes that AI can be an ‘enabler’ for 134 of the 169 targets within the SDGs. That’s a 79% hit rate, but any optimism should be cautioned by its other claim that 59 of the targets (or 35%) may experience negative effects from AI, such as a widening of inequalities. For SDG4 in particular, AI can be an enabler of all 10 targets but a threat to 7 of those targets. History tells us that technology is a double-edged sword in education; with AI the blade has never cut so deep. AI can catalyse progress towards SDG4 by enabling learning experiences that are tailored to each learner’s needs. The same technologies, though, threaten to widen the attainment gap by benefiting communities that already possess the resources and expertise to make use of such tools. Equalisation of opportunity requires investment in infrastructure and capacity building at the ground level so that the most vulnerable, resource-starved communities can benefit from learning technologies.

AI is allied with big data, and those who embrace these technologies have to contend with the issue of privacy. There are stark examples already of facial recognition software being applied in the classroom, ostensibly to track students’ emotions in real time. Such technologies raise all kinds of ethical issues, ranging from consent to the reliability of automated judgements. Left unchecked, these tools may deservedly find themselves on the next decade’s list of EdTech debacles. Regulatory oversight of AI is a live wire across many industries, though it is given relatively little attention in education circles, as our interview with AI expert Wayne Holmes reveals.

 

 

Upskilling tomorrow’s workforce

The advent of AI is also exerting untold pressure on the workforce, which surely has ramifications for how we define quality learning outcomes. According to one OECD study, based on the Program for the International Assessment of Adult Competencies (PIACC), ‘automation risk’ is higher in countries with a lower GDP per head. It stands to reason: routine cognitive skills are precisely the ones that machines can most easily replicate, and these skills are more prevalent in the workforces of poorer countries. The study makes the startling claim that, in fact, “There are no examples of education systems that prepare the vast majority of adults to perform better in the three PIAAC skills areas than the level that computers are close to reproducing.” That is, even the highest-performing education systems are failing to equip students with the literacy, numeracy and problem-solving skills that their future jobs will rely on.

SDG4 risks losing its relevancy. Specifying a commitment to quality outcomes was indeed a quantum leap forward, but we must also evaluate and re-evaluate all that ‘quality’ must entail in 2030. Proficiency in core literacies is a lower bound: a minimal level of attainment we should imbue all students with. But quality education must represent more than routine, easily automated skills. It must also have room for ‘non-routine’ skills. The challenge, of course, is that such skills are less straightforward to define, let alone measure. But that cannot be a reason to abandon them altogether, and the OECD’s PISA 2021 Mathematics Framework is just one example of how some educators are envisioning a curriculum fit for our times.

 

 

‘Quality’ education in the age of AI

There is a risk, I believe, of lurching towards an extreme of removing core proficiencies altogether. It remains essential, even in this age of high-powered computers, for students to acquire fluency in core literacies such as reading and numeracy. But just as important is the ability to apply those core skills in novel ways, to find meaningful connections between ideas and to develop flexible understanding of concepts. In a more optimistic review of the past decade, prominent maths educator Dan Meyer highlights the potential of technology to bring such non-routine skills to life in mathematics. “Math education needs visualizations that provoke students to wonder mathematically,” he says. “It needs a creative palette that enables students to express their mathematical ideas more fully. It needs to connect ideas and people together so that students and teachers can learn from each other’s mathematical creativity. Computers are great at the right tasks too: visualization, creation, and connection. Let’s put them to work.”

I would argue that AI technologies need to be put to work to ensure all students are fluent in the core literacies, but also to enable the deeper learning experiences Meyer speaks of.

The SDGs were moulded during a period of rapid technological progress. They make implicit assumptions around what constitutes a quality education, assumptions that are being challenged by every advance in AI. Our notions of education cannot remain static, for it would be a cruel irony to achieve SDG4, only to realise it has failed to prepare students with the knowledge, skills and dispositions they need to thrive in 2030. Quality learning outcomes must, as part of their definition, represent the outcomes most relevant to tomorrow’s world.