Before artificial intelligence: automation, calculation and classification in the history of ideas.

Euclid and repeatable procedure

Book VII of the Elements describes a procedure for finding the greatest common divisor through defined and repeatable steps.

Modern connection: a precise, finite and repeatable procedure is the basis of an algorithm.

The sieve of Eratosthenes

The procedure identifies prime numbers by progressively eliminating those that do not satisfy the condition.

Modern connection: filtering invalid cases resembles workflow rules, filters and checks.

Aristotle: categories and inference

Aristotelian logic organises forms of inference; the Categories provides a structure for distinguishing general kinds of beings and predication.

Modern connection: classifying concepts, representing rules and separating premises from conclusions.

Does correctly applying a rule amount to understanding?

The Antikythera mechanism

A complex geared device represented calendrical and astronomical information and performed calculations relating to celestial cycles.

Modern connection: a model of the world can be embedded in a device that transforms initial settings into a calculated result.

It did not learn from data and was not artificial intelligence: it executed relationships designed by people.

Hero of Alexandria and automata

The Automata describes mechanical devices capable of executing sequences designed in advance, including mobile automata and theatrical displays.

Modern connection: a prepared sequence is executed automatically in the same order, a useful analogy for automation but not machine learning.

Al-Khwārizmī and the name of the method

The Persian mathematician sets out systematic procedures for calculation with positional notation. The Latin form of his name, Algoritmi, enters European usage and becomes the word we still use.

Modern link: the word "algorithm" comes not from an abstract theory but from a working manual.

Leibniz: binary and the calculus of reasoning

Leibniz formalises the binary number system and imagines a calculus ratiocinator: a notation that would settle disputes by calculation rather than argument.

Modern link: two separate ideas that now live inside every machine: binary representation, and the ambition to reduce reasoning to mechanical operation.

Boole and the laws of thought

An Investigation of the Laws of Thought treats propositions as algebraic objects: true and false become values manipulated by explicit rules.

Modern link: every condition, filter and gate in an automated sequence is Boolean algebra in use.

Jacquard, Babbage and Lovelace

The Jacquard loom separates the machine from its instructions, punched onto cards. Babbage designs the Analytical Engine; Ada Lovelace observes it could operate on any symbols, not only numbers.

Modern link: the split between machine and program, and the insight that computation is not only about quantity.

Turing: computability and a behavioural criterion

Turing defines what a machine can compute in principle, and in 1950 moves the question "can it think?" onto observable ground, proposing an imitation game in place of a definition.

Modern link: the boundary of what is computable, and the habit of judging systems by measurable behaviour rather than stated intent.

Shannon and the measure of information

A Mathematical Theory of Communication separates content from meaning: information becomes a measurable quantity, with noise, redundancy and channel capacity.

Modern link: why data can be abundant and uninformative, and why quality checks measure noise before signal.

Dartmouth: the term is coined

The proposal for the Dartmouth summer workshop introduces the phrase "artificial intelligence" and sets a research agenda. From here on the history stops being analogy and becomes that of the field itself.

Modern link: this is where the page stops: everything after it is no longer a conceptual antecedent but the direct history of the field.