What “ancestral intelligence” means when Silicon Valley says AI — and why 256 Odu belong in that sentence

The proof: the phrase is already in the AI conversation

In April 2025, policy researcher Nanjira Sambuli gave a TED talk titled What ancestral intelligence can teach us about AI. TED's own page for the talk frames it this way: there is a common African proverb, "When elephants fight, it's the grass that suffers." Sambuli applies that thinking to today's AI evolution and asks: when tech giants battle for dominance, who gets trampled in the process? She introduces what TED calls a new ethical compass for AI, and shows how people across the continent are charting a different path for the future of tech.

Read the source yourself: TED: What ancestral intelligence can teach us about AI

That is a policy and ethics argument. It is about who gets harmed, who gets a say, and what values should steer the technology. It is a serious use of the phrase "ancestral intelligence," and it is not the subject of the book discussed below.

The damaging admission

Sharing a phrase is not sharing a stage. The Original AI: Ancestral Intelligence is not endorsed by TED, is not an official companion to Sambuli's talk, and is not a white paper from any government or technology company. Sambuli did not write it or review it, and this article does not suggest she knows of it. The only link between the talk and the book is that both use the words "ancestral intelligence" and both are part of a larger conversation about where intelligence comes from.

That is worth saying plainly because it is the first thing a careful reader should check, and because everything else in this article is stronger when that line is clean.

Who this is for, and who it is not

This is for you if you keep hearing "African AI," "AI ethics," or "Ifa as code" and want a book that makes a specific, testable-in-the-open argument rather than a slogan. It is for readers who already know a little about Ifa, or who are curious about the history of binary systems, and want a long-form case from inside the tradition.

This is not for you if you want a peer-reviewed computer science paper, a TED-affiliated resource, a guarantee of enlightenment, or an audiobook. The book is an ebook, and nothing here claims otherwise.

The moment the slogan stops

Picture the feed after another foundation-model launch. The ethics threads roll in. A panel on "African AI" gets clipped. Somebody comments, "We had intelligence systems before Silicon Valley." Fifty people like it. Then the thread ends, because nobody has a next sentence.

The book exists for the next sentence. Sambuli's compass asks who gets harmed when giants fight. The book asks a different but connected question: what did the older intelligence systems actually look like, and is Ifa one of them? Its answer, offered as an argument and not as settled consensus, is yes, and it points to the 256 Odu of Ifa.

The arithmetic nobody disputes

Start with the part that is just math. In Ifa divination, a priest (a Babalawo) generates a figure called an Odu through repeated binary outcomes. Using the opele divining chain or sacred palm nuts, each throw produces one of two results: a single mark or a double mark. Repeat that across eight positions and the number of possible outcomes is two to the eighth power, which is 256.

That is where the 256 Odu come from. It is also, not by coincidence in the book's view, the number of values in a computing byte: eight bits, 256 combinations, from 00000000 to 11111111. The book notes that the first Odu, Ogbe, can be written as eight single marks (all ones), and that Oyeku can be written as eight double marks (all zeros), with the remaining 254 Odu sitting between those poles. It also notes that the Chinese I Ching works on six positions, giving 64 hexagrams, where Ifa works on eight.

None of that arithmetic is controversial. Eight binary choices give 256 states. The question the book raises is what to make of the correspondence.

What the book argues, and where the argument begins

The book's central claim, stated in its own technical chapter, is that Ifa is not just structurally similar to an information system but should be understood as one: an intelligence system built on binary combinatorial logic, a large body of case-based verses, trained human specialists, and a practice of producing actionable guidance from ambiguous input.

It lays out the pieces like this:

• The address space. 256 Odu, organized as 16 principal Odu and 240 derived combinations.
• The corpus. Each Odu carries a body of verses (ese Ifa), each containing a story, a diagnosis, and a prescribed action. A trained Babalawo memorizes thousands of these.
• The practitioner as processor. The book describes the Babalawo as a trained associative memory: the divination result is the input, the verses are the stored knowledge, and the interpretation matched to the person in front of them is the output.
• Training and feedback. Years of supervised study, correction by senior practitioners, and a community that rewards accurate counsel and withdraws trust from poor counsel. The book compares this to pre-training, fine-tuning, and feedback in modern machine learning.

This is the book's case, and it is a case, not a consensus. The comparison to neural networks and language models is an analogy the author chooses to press hard, and many readers, including many scholars of Ifa and many computer scientists, will want to weigh it carefully. The article you are reading does not ask you to accept it. It asks you to read the argument in its own words and test it.

For independent context, there is outside recognition of the system's complexity. UNESCO lists the Ifa divination system on its Representative List of the Intangible Cultural Heritage of Humanity, and describes it as making use of an extensive corpus of texts and mathematical formulas. That is a heritage designation, not an endorsement of any computing claim, but it is a useful, neutral anchor for the idea that Ifa is a structured knowledge system. See the UNESCO listing for the Ifa divination system.

The history claim: from Ifa to the byte

The book then makes a historical argument that is more contested than the arithmetic. It traces binary mathematics through the I Ching and Gottfried Leibniz, who published an essay on binary arithmetic in 1703 that acknowledged the I Ching's hexagram structure as a predecessor, and proposes that Ifa's binary structure may sit upstream of that line of transmission.

Be careful with that sentence. The first half, that Leibniz formalized binary arithmetic and connected it to the I Ching, is well documented in the history of mathematics. The second half, that there is a traceable line running from Ifa to the I Ching and on to modern computing, is the book's thesis. The book presents it as an argument supported by the evidence it assembles, and it argues that the burden of proof sits with those who dismiss the parallel. Readers who want to push back should. A good argument survives pushback, and a good reader tests it.

Ethics as architecture

Here is where the book speaks most directly to the conversation Sambuli opened. In the book's account, modern AI treats ethics as a layer added after the fact: filters, guardrails, and fine-tuning bolted onto a base model. Ifa, the book argues, treats ethics as the core structure. The guiding principle in every consultation is iwa pele, gentle and measured character, and training is aimed at forming the practitioner so that harmful guidance is not just forbidden but unthinkable.

The book also argues that context is the medium of Ifa. A language model processes your words. A Babalawo, the book says, responds to you: your history, family, health, and alignment between the destiny you chose and the road you are on. It sets "response to the person" against "statistically likely reply to a prompt."

You can disagree with the framing and still find the questions useful. Who is accountable for the output? What does the system know about the person it is advising? Is ethics designed into the structure or applied afterward? Those are live questions in AI policy today, and they are the same family of questions behind the phrase "ancestral intelligence" in the first place.

Ori: the intelligence closest to you

The book also gives space to Ori, the Yoruba concept of the personal head or destiny, and its chapter title makes a bold claim: "Ori: The Original Artificial Intelligence Is You." In this reading, the most important intelligence in the system is not the oracle or the priest but the self whose destiny is being read. For readers coming from an AI-ethics direction, this is where the book turns from structure to the individual: the system serves a person's alignment with their own path.

What you will find inside

The table of contents lays out the shape of the book:

• Part One: The Argument. The 256 Odu as a source code, binary mathematics and the I Ching, what the book calls the suppression of that knowledge, and Ori.
• Part Two: The 16 Principal Odu. One chapter per principal Odu, from Ogbe through Ofun, each framed as a type of intelligence, such as endings and transformation, inner vision, boundaries and self-protection, or healing and medicine.
• Part Three: The Complete Source Code. A reference chapter on the 240 minor Odu, the technical argument for Ifa as an AI system, practical guidance for working with the Odu, and a closing chapter on reclaiming ancestral intelligence.

It also includes a glossary of Ifa and Yoruba terms, which helps if you are coming to this from the technology side and the vocabulary is new.

What it will not do

The book will not make you enlightened. It will not give you a trading edge, a startup idea with guaranteed returns, or a verified result that Ifa predicts anything. It will not settle the debate about whether divination should be classed as computation. It is a long argument from within a tradition, written to persuade, and it should be read as one.

It also will not replace the people who carry Ifa. If the book moves you toward the tradition, the tradition has teachers, Babalawo and Iyanifa, and communities who can tell you what a book cannot.

Reason why now

The reason to publish this today is not a sales number or a ranking, and none is claimed. It is the timing of the conversation. "Ancestral intelligence" has entered public AI discourse through a respected TED talk, and a reader who hears the phrase and wants a deep, primary-source-style treatment of one specific ancestral system, Ifa and its 256 Odu, has a title written for exactly that question and available on both major ebook stores.

How to test it for free

Do the arithmetic yourself: two choices, eight positions, how many outcomes? Look up the UNESCO page. Watch the TED talk and write down Sambuli's questions. Then ask whether the book's claims about Ifa as a knowledge system help you answer them or complicate them. If you want the full argument, the Odu chapters, and the glossary, the book is on the stores below. Read the description and sample first, and judge it on the merits.

Ase.

Get the book

• The Original AI: Ancestral Intelligence on Google Play
• The Original AI: Ancestral Intelligence on Apple Books
• Back to the Robert Shumake hub

Get the next Living Archive / book brief before the feed buries it

The list is author-owned. No spam. One click to leave. Add your email in the signup form directly below this article, then come back to the book on Google Play or Apple Books whenever you are ready.