Egyptian Coder Builds AI Model to Represent His Culture

When Egyptian coder Assem Sabry searched for an AI model that truly understood his culture, he came up empty. So he did what any determined innovator would do: he built one himself.

Sabry named his creation Horus, after the ancient Egyptian god of the sky. His goal was simple but powerful: create an AI that speaks to Egyptians in their own cultural context, not through the lens of American or Chinese tech giants.

Using GPUs from Google Colab and open-source datasets, Sabry trained Horus from the ground up. When he released it in early April, more than 800 people downloaded it in just the first week.

He's not alone in this mission. Developers around the world are tackling a massive imbalance in artificial intelligence: most AI models work brilliantly in English and reasonably well in Chinese, but they struggle with almost everything else.

The irony is striking. Languages labeled as "minority" are actually spoken by the global majority. But because AI models learn from massive web scrapes, and because English dominates the internet, other languages get left behind.

In 2023, researcher Aliya Bhatia published a study showing how commercial pressures pushed Big Tech to prioritize English. Companies rushed to capitalize on AI without addressing the language gap, partly because training data for other languages was limited.

For years, the economics reinforced this cycle. Training AI costs serious money, and companies saw little reason to invest in smaller language communities without guaranteed profits.

The Ripple Effect

Now that's changing. The rise of open-source AI models has opened doors that were previously locked. Two years ago, Sabry notes, the technology wasn't good enough and wasn't available to everyone.

Today, developers worldwide can build their own models from scratch. This democratization means AI can finally serve communities that tech giants overlooked, preserving cultural nuances and linguistic diversity that would otherwise vanish in translation.

Challenges remain, Bhatia points out. Computing power, infrastructure, and funding still create barriers for many developers. But the momentum is undeniable.

Sabry's success with Horus shows what's possible when innovation meets determination, and when technology finally starts speaking everyone's language.