It Takes a Village to Raise AGI

There is a well-known African proverb: it takes a village to raise a child. Growing up, I lived this reality every day. If a neighbor caught you swearing at another kid, they wouldn't wait to ask your parents' permission — they would discipline you right there. Conversely, you could leave your house in the morning without a single worry about where you would eat lunch. If lunchtime found you at a friend's house, you ate with their family, no questions asked. The community shared the responsibility of raising you, correcting your blind spots, and providing what you needed to grow.

Today, we are standing at the precipice of Artificial General Intelligence (AGI) and Artificial Superintelligence (ASI). And just like a child, AGI cannot be raised in isolation. If it is going to serve all of humanity, it literally takes a global village to raise it.

Section 01The Rumsfeld Trilemma and AI's Blind Spots

Despite their brilliance, today's frontier AI systems have massive foundational gaps. This brings to mind the famous Donald Rumsfeld trilemma: there are known knowns, known unknowns, and the most dangerous of all, unknown unknowns.

Imagine a scenario in the near future. An AI lab proudly announces they have reached AGI. Somewhere in the world, a five-year-old child — one of over 400 million Bantu speakers — sits down with this AGI and asks for help learning to read the word ulebomba.

The AGI, relying on how it was trained, begins by decomposing the sentence morphologically. It breaks the word down by its grammatical parts. But there is a glaring problem: that is not how a Bantu child learns to read, write, or pronounce sounds. They learn syllabically.

The AGI fails the five-year-old because it has encountered an unknown unknown. Foundationally, it does not know the full syllable inventory of the language. In fact, presently, not a single frontier model can tell you the full syllable inventory of any of the 500+ Bantu languages.

Section 02The Missing Periodic Table

For Bantu languages, the syllable set is not just a grammatical quirk — it is the absolute foundation.

You cannot write a single English word without using the 26-letter alphabet. You cannot build life without the 20 amino acids. You cannot do chemistry without the periodic table. For a Bantu language, the full set of syllables is its periodic table. It is the true alphabet. Every Bantu person who has ever learned to read and write learned this at the foundation level.

Yet, the systems claiming to be approaching AGI are missing this absolute fundamental truth. They are trying to build the roof without pouring the concrete.

Section 03Bringing Our Share to the Village

When I say it takes a village to raise AGI, I mean it literally. We each have a responsibility to recognize the shortcomings in these systems, and where we have the power to fix them, we must step up and do the work. The labs in Silicon Valley cannot build this alone; they do not have the cultural or linguistic context.

This is why, at BantuNomics, we have curated over 400 Full Syllable Inventories for Bantu languages.

We do not view this simply as curating a dataset. This is infrastructure. We are laying down the foundational linguistic roads and bridges required to ensure that when AGI finally arrives, it does not leave 400 million people behind. We are building the tools to ensure that a Bantu child can ask a frontier model for help, and the model will understand them entirely on their own terms.

AGI will only truly be "general" when the whole village contributes to its upbringing. At BantuNomics — and through amina, where the village literally gathers to record what AI needs to hear — we are building our part.