Is machine learning a form of divination?

No. They share a structure — a corpus, an opaque procedure, and a reading a person acts on — but a model generates novel output statistically, while divination retrieves from a fixed corpus and requires an accountable human interpreter.

What do Ifá and the I Ching have in common with binary computing?

Both encode outcomes in base-two marks: the I Ching in six broken or unbroken lines producing 64 hexagrams, Ifá in eight marks producing 256 Odu. The encoding is binary; the systems themselves are interpretive, not computational.

What does this comparison get wrong if pushed too far?

It fails if used to claim equivalence. Divination has bounded output and a named interpreter answerable to the client; a model has neither, and treating machine output with oracular deference removes the interpretive step entirely.

Divination systems and machine learning: what actually rhymes

Divination systems and machine learning both take an underdetermined situation and return a structured reading a person then acts on. Both operate over a symbol space, both rely on a procedure the user does not fully inspect, and both derive authority from a corpus rather than from the practitioner's individual reasoning. They differ fundamentally in that a divination corpus is fixed and interpreted by an accountable human, while a model generates novel output with no interpreter obligated to the person asking.

The genuine parallels

Both produce output from a procedure the person consulting it cannot audit. Both compress prior knowledge into a form retrievable at speed. Both are consulted precisely at moments of uncertainty, where the person needs a structure to think with rather than a fact to verify.

And both derive their authority from a corpus. The practitioner's credibility rests on the tradition; the model's rests on the training set. Neither asks the user to accept the output because the operator personally reasoned it out.

The differences that matter

A divination corpus is finite, enumerated, and stable. A model's is vast, mutable, and largely undisclosed. Ifá retrieves; a model generates, which means it can produce a confident answer that exists nowhere in its sources.

Most importantly there is a named interpreter in the traditional system, embedded in a lineage, answerable to the client and the community. Machine output arrives without that accountability, which is the practical reason the comparison should not be used to launder either system's authority.

Why run the comparison at all

Because it makes the interpretive layer visible. Traditional systems are explicit that a reading requires a trained interpreter and a situation; machine output is presented as if it needs neither. The comparison restores that missing step.

Read the books behind this

The Original AI: Ancestral Intelligence

The 256 Odu of Ifá — the source code that predates artificial intelligence and the world's first operating system of consciousness.

Get the book Apple Books

Questions and answers

Is machine learning a form of divination?: No. They share a structure — a corpus, an opaque procedure, and a reading a person acts on — but a model generates novel output statistically, while divination retrieves from a fixed corpus and requires an accountable human interpreter.
What do Ifá and the I Ching have in common with binary computing?: Both encode outcomes in base-two marks: the I Ching in six broken or unbroken lines producing 64 hexagrams, Ifá in eight marks producing 256 Odu. The encoding is binary; the systems themselves are interpretive, not computational.
What does this comparison get wrong if pushed too far?: It fails if used to claim equivalence. Divination has bounded output and a named interpreter answerable to the client; a model has neither, and treating machine output with oracular deference removes the interpretive step entirely.