Part III · Era 1 — Precursors (until 1955)

draft

Before an electronic computer existed, people already dreamed of mechanized reasoning. This era is the prehistory of AI: formal logic, the theory of computation and cybernetics prepare the conceptual ground.

Narrative

The question "can a machine think?" is ancient, but it only gains rigorous footing when reasoning is reduced to symbol manipulation (logic) and when it is defined what a machine can compute (Turing). In parallel, the idea that the brain is a network of processing units (McCulloch-Pitts) and that systems self-regulate through feedback (Wiener) provides the biological metaphor and the vocabulary of control.

Milestones

  • ~350 BC — Syllogism · Aristotle · first formal theory of valid inference.
  • 1843 — Notes on the Analytical Engine · Ada Lovelace · envisions that

    machines could manipulate symbols beyond numbers.

  • 1847 — *The Laws of Thought* · George Boole · the algebra of logic — basis

    of every digital circuit.

  • 1879 — Predicate logic · Gottlob Frege · formalizes quantification and

    proof.

  • 1936 — Turing machine · Alan Turing · defines computability; the abstract

    model of every computer.

  • 1943 — Artificial neuron · McCulloch & Pitts · first mathematical model of

    a neuron; logical networks.

  • 1948 — *Cybernetics* · Norbert Wiener · control and communication through

    feedback in animals and machines.

  • 1948 — Information Theory · Claude Shannon · quantifies information (bit,

    entropy) — underpins everything that comes after.

  • 1949 — Hebbian learning · Donald Hebb · "neurons that fire together wire

    together" — basis of plasticity and learning in networks.

  • 1950 — *Computing Machinery and Intelligence* · Alan Turing · proposes the

    Turing Test and debates the "thinking machine".

People

Alan Turing (1912–1954) — father of computer science and theoretical AI. Computability (1936), Turing Test (1950), cryptanalysis at Bletchley Park. 🖼️ Photo F-III.1 — photographic portrait (commonly public domain).

Claude Shannon (1916–2001) — father of information theory. Bit, entropy, channel capacity; also a pioneer of computer chess (1950). 🖼️ Photo F-III.2.

Warren McCulloch & Walter Pitts — the first mathematical neuron (1943). They showed that networks of logical neurons can compute any propositional function. 🎨 illustrated portrait if no free photo exists.

Norbert Wiener (1894–1964) — founder of cybernetics. Feedback and control — vocabulary that AI and RL still use.

Curiosities

  • Turing's lost BBC broadcast (1951). Turing gave a radio talk on thinking

    machines for the BBC, part of a series featuring other pioneers of the emerging science of computing. No recording of the transmission survives — but the original script, with his handwritten notes and markings, is kept in the BBC Written Archive.

  • A prediction with a deadline. In the 1950 paper Turing bet that within about

    50 years (i.e. by ~2000) computers would store enough that an interrogator would have no more than a 70% chance of telling human from machine after five minutes of questioning.

  • Machine learning, foreseen. Rather than programming everything up front,

    Turing proposed building systems that learn from data, comparing a child's brain to "a notebook one buys from a stationer's: rather little mechanism and lots of blank sheets."

  • Persecution. Turing's cryptanalysis is credited with shortening WWII, yet in

    1952 he was convicted of "gross indecency" and subjected to chemical castration; he was found dead of cyanide poisoning in 1954, ruled a suicide.

Source: curiosities surfaced by the BBC News Brasil report *"Máquinas podem pensar? Os 70 anos de história que levaram à inteligência artificial dos dias atuais"* (Luiz Fernando Toledo, 2025), cross-checked against primary sources.

🎨 Figure F-III.3 — From logic to the machine. Brief: conceptual illustration: on the left, logical symbols and Boole's algebra carved into stone/paper; in the center, the tape of the Turing machine; on the right, a stylized McCulloch-Pitts neuron turning into a circuit node. Shows the "passage" from abstract thought to mechanism. Compendium palette.

Connection to the Compendium: this era is the source of the formal inputs that Part I maps (logic, information theory, computability) and of the inspirations (neuron, feedback) discussed in ../01-life-cycle/04-nature-of-inputs.kmd.