When AI Promises the World But Can’t Cross the Educational Divide

“To those who have, more will be given.”

This biblical phrase, which sociologist Robert K. Merton dubbed the Matthew Effect, has long haunted discussions of inequality. As I watch the AI hype, particularly Large Language Models (LLMs), and their supposed effects education, I can’t help but wonder: are we on the brink of another example of the Matthew Effect?

This blog has been in my drafts for about a month, because it might be a bit controversial. I came to this understanding as I was reading for my Culture, Heritage and Learning Futures course.

The promise of AI in education might seem enticing. Tech has often proposed itself as the solution to educations problems, and I’ve heard many times people propose AI ‘solutions’ to education “to reduce inequalities.”

But let’s look at the reality on the ground. In many parts of the world, soe of which I experienced first hand, schools struggle with basic infrastructure – reliable electricity, internet connectivity, or even adequate classroom space. How can AI tools help when students can’t even access them? The gap between those who have access to these technologies and those who don’t isn’t just widening – it’s becoming a chasm.

Then there’s the differences in the data that these models are trained on. Most LLMs are trained on data from the Global North, predominantly in English. What about the knowledge, perspectives, and experiences of the Global South? What about the rich cultural and educational traditions that exist in non-English speaking communities? These AI systems, trained on a narrow slice of human knowledge and experience, risk further marginalizing already underrepresented voices and ways of knowing.

The language barrier adds another layer of complexity. While there’s much excitement about AI’s potential for personalized learning, the reality is that most cutting-edge AI educational tools are developed in English, for English speakers. Students who speak other languages often get second-rate translations or nothing at all. Even when translations exist, they often miss cultural nuances and context that are crucial for genuine understanding.

Consider a student in rural Asia. They might have occasional access to a shared smartphone, intermittent internet, and English might be their third or fourth language. How can they compete with a student in who has personal devices, high-speed internet, and native English fluency? The AI revolution in education threatens to leave them even further behind.

Some argue that AI will eventually bridge these gaps, that costs will come down and access will improve. But history suggests otherwise. New technologies tend to benefit first – and most – those who already have advantages. By the time they “trickle down” to underserved communities, new technological gaps have already emerged.

Here’s the uncomfortable truth: AI won’t fix educational inequality because AI can’t address the systemic and structural inequalities that create educational disparities in the first place. Technology alone can’t solve problems rooted in poverty, discrimination, and historical injustice.

The solution to educational inequality isn’t AI – it’s us. It’s about political will, resource allocation, and a genuine commitment to educational equity. It’s about ensuring every school has adequate funding, every teacher has proper training, and every student has the basic resources they need to learn. Only then can we talk meaningfully about how AI might enhance education for everyone, not just the privileged few.

The real work of addressing educational inequality requires human commitment, political action, and systemic change. To prevent the Matthew Effect from amplifying existing inequalities, we need to focus less on technological quick fixes and more on addressing the fundamental inequities in our educational systems.

Remember: AI won’t address inequality. We have to.

[Featured image generated by Adobe Photoshop AI]