Once upon a time, the world looked to education as humanity’s great equaliser. Much has been said about the power of education to improve people’s lives, overcome social differences, and lift whole communities out of poverty. On a larger scale, it plays a key role in driving progress, advancing sustainable development, and even promoting peace.
But it’s not that simple, because for education to have a widespread positive impact, it must be accessible to everyone regardless of their circumstances. Since this is not the case today, education has turned from an equaliser into one of our most enduring fault lines.
While industrialised nations enjoy unprecedented connectivity and innovation, school attendance remains out of reach for 272 million children in developing countries, as per UNESCO’s latest records. Alarmingly, this number has reduced by less than 1% since the UN designated quality education as its fourth sustainable development goal (SDG) in 2015, even though it is a key enabler of all other SDGs.
Technology to the rescue
New technological advancements, especially in the field of AI, have sparked hopes of bridging this gap. AI is already bringing immense advantages to education, including innovating teaching methods, customising learning experiences, and freeing up teachers from mundane admin tasks to focus on what’s most important – shaping young minds.
Education has turned from an equaliser into one of our most enduring fault lines
But with the world’s attention locked on the rapid evolution of AI, the race to build ever-larger models is creating a digital divide that mirrors existing inequities in global education. While billion-parameter systems may dazzle, they are largely inaccessible to educators and learners in under-resourced settings, both technologically and academically. This disparity not only limits the reach of advanced edtech but also exacerbates the challenges faced by those already marginalised.
Small, efficient AI models win the race
To ensure that the benefits of AI in education are equitably distributed, its future must be guided not by scale alone but by access, inclusion, and human-centred learning outcomes. Luckily, recent shifts in research are pointing in this direction. Developers are increasingly turning to small, efficient models that mimic how humans learn language, aligning with efforts to close the global learning gap by enabling AI to redefine the learning journey of every child.
Crucially, the effectiveness of smaller models depends not on volume, but on quality. High-quality training data, including well-curated educational content, culturally relevant material, and age-appropriate pedagogy, is essential. Because smaller models can be trained with more compact datasets, making them faster and more cost-effective to build, they are also far more sensitive to the quality of the data they learn from. When trained responsibly, with rigorous attention to content accuracy and learning outcomes, these models can deliver high-impact educational support without the excessive resource demands of large-scale systems.
Smaller models can be designed to understand and respond to local languages, cultures, and educational needs. And unlike larger, resource-intensive AI systems, smaller models also require less computational power. This makes them more affordable and easier to implement in low-resource settings, where traditional educational tools may be prohibitively expensive. By lowering the barrier to entry, these models can be deployed widely, reaching underserved populations.
The implementation of smaller AI models provides fertile ground for synergies among local stakeholders – from teachers to community leaders. These collaborations foster a sense of ownership and ensure that educational solutions are grounded in the realities of the communities to which they cater.
As education systems worldwide face shrinking budgets and rising expectations, smaller, smarter models could be the key to scalable, affordable, and locally adaptable edtech solutions that are accessible to everyone, unlocking global learning equity.
AI should support educators, not operate unchecked, and its application must always align with ethical standards and educational values
However, technology alone is not enough. The responsible deployment of AI in education must be underpinned by strong governance frameworks and meaningful human oversight. Responsible AI requires transparency in how systems are developed and used, safeguards to prevent bias, and clear accountability for outcomes. AI should support educators, not operate unchecked, and its application must always align with ethical standards and educational values.
To realise this vision, a multifaceted approach is essential. First, digital literacy programmes must equip educators and students with the skills needed to understand and use AI tools effectively and critically. Second, AI-powered tutoring solutions should offer personalised learning experiences that adapt to individual needs while remaining aligned with curriculum standards and pedagogical best practice. Third, sustained investment is needed in AI infrastructure and initiatives that explicitly target underserved communities, ensuring that innovation does not deepen existing divides.
AI – a helpful tool, not a teacher
Some may ask: With all its advanced capabilities, will AI ever replace teachers? The truth is, in an age of extreme technology, schools need the best humans more than ever. While technology is undoubtedly a helpful tool, it can never replace the human touch. It can’t satisfy children’s emotional and social needs, it can’t instil strong moral values in them, and it certainly can’t be a role model.
Eventually, Al and robots will become a commodity, with teachers taking on the role of ambassadors and mentors for human beings – ‘human mentors’. Al will, in fact, enhance the value that humans can bring to education. And creating value for humans is one of the most powerful competitive advantages you can have in education.
Building an AI-powered future for every learner
Educators, technologists, and policymakers must work together to build an AI future that benefits the hundreds of millions of children still out of school today, not just the well-connected few. Only with collective efforts can we harness the transformative potential of AI to put quality education within the reach of every learner. This way, education can become a universal empowerment tool and a true catalyst of stability, prosperity, and peace.
