Back to Insights

AI • Language • Digital Divide

AI spoke English. Billions didn't.

FG
Felix Ghauri

· 3 min read

AI spoke English. Billions didn't.

AI spoke English. Billions didn’t. Now that barrier may just have fallen

In under three years, 1.2 billion people used AI tools. Faster than the internet, the PC or the smartphone.

Look at this map from Microsoft’s AI Diffusion Report: the UAE hits 59.4%. Singapore 58.6%. Norway 45.3%.

Then scan across Africa, South America, Central Asia. Red zones. Single digits. Regions that are home to billions.

Until now, English has dominated. Models were trained on English data and served English speakers first.

Meta just cut that constraint. Its ‘Omnilingual ASR’ recognises speech across 500 languages and can learn new ones from small local samples. A few voice clips can unlock a language.

But here’s the gap: technology moved at exponential speed. Infrastructure, literacy, connectivity didn’t.

The language problem is solved, but the access problem remains.

When AI finally speaks your language but still can’t reach you, what exactly did we solve?

💬 Join the conversation on LinkedIn

View on LinkedIn →
FG

Felix Ghauri

Applied AI Practitioner · Founder, Futures Forum

Felix helps organisations navigate AI and exponential change. He writes about technology, geopolitics, and the future of work.

Thinking about AI in your workflow?

Let's discuss what might work for you.

Let's Talk