Long-form writing by Munyambala on Bantu phonology, frontier-AI capability gaps, infrastructure framing, and the shape of the work being done at 3Mega's two initiatives — BantuNomics and amina.
When an AI lab encounters a Full Syllable Inventory for the first time, the question they ask is "what is the licensing fee." That is the wrong frame. FSIs are not a dataset you train on. They are the substrate every downstream operation references — closed, enumerable, standardised. Subscribe accordingly.
The syllable is the unit. Not the letter. By age 6 a Bantu child has internalised the syllabary of their language — the same inventory a frontier LLM in 2026 cannot enumerate. One layer off, and the error propagates through every prediction.
Tone in a Bantu language is not decoration — it is meaning. And tone does not attach to the letter or the morpheme. It attaches to the syllable. If a frontier model cannot enumerate the syllables of a Bantu language, it cannot understand its tone.
Bantu languages have been spoken for thousands of years. The writing system is barely a hundred years old — and it leaves out the part that decides meaning. We named this gap so the AI labs could see it.
There is a well-known African proverb: it takes a village to raise a child. AGI is no different. If it is going to serve all of humanity, it literally takes a global village to raise it.