Trail10 steps

Thinking machines: how AI began

From bronze giants and reasoning wheels to a summer workshop, a chatbot therapist, two winters and a bedroom GPU.

Artificial intelligence didn’t start with chatbots. It started with an old philosophical hope, that thinking could be done by rules, and took three centuries, a world war, two funding collapses and a lot of hype to become what it is now. Ten stops, in order.

Start the trail →

I
It starts with a question older than any machine: could reasoning itself be mechanised?
No. 002ConceptArtificial Intelligence
The dream of thinking machines is thousands of years old Long before computers, people imagined bronze giants, reasoning wheels and a language in which arguments could be settled by calculation.
II
In 1837 Babbage designed a machine that could, in principle, follow any instructions.
No. 004ObjectTechnology
The computer that was designed in 1837 Charles Babbage designed a steam-age machine with memory, a processor and punched-card programs, then never finished building it.
III
Ada Lovelace saw what it could become, and doubted it could ever truly originate anything.
No. 001PersonTechnology
The first computer programmer was a poet’s daughter In 1843, a century before electronic computers, Ada Lovelace wrote out a program and imagined machines making music.
IV
Meanwhile, audiences were happy to believe in a fake thinking machine.
No. 080ObjectArtificial Intelligence
The chess-playing machine that had a man inside From 1770, a turbaned mechanical figure beat players across Europe and America at chess. It was a brilliant hoax, and Amazon named a service after it.
V
In 1950 Turing asked the question directly, and answered Lovelace by name.
No. 123PersonArtificial Intelligence
“Can machines think?”: Turing’s 1950 question Alan Turing opened a 1950 paper with that question, then replaced it with a test, and answered Ada Lovelace by name.
VI
Six years later, a summer workshop named the field and promised fast progress.
No. 032EventArtificial Intelligence
AI got its name from a summer workshop proposal In 1955 four researchers asked for funding for “a 2-month, 10-man study of artificial intelligence”, confident they could make a significant advance in one summer.
VII
The first learning machine arrived with headlines about consciousness.
No. 094StoryArtificial Intelligence
In 1958 the Navy said its neural network would one day be conscious The first trainable neural network, the perceptron, was announced as the embryo of a computer that would “walk, talk, see, write, reproduce itself and be conscious of its existence”.
VIII
A simple chatbot showed how easily people see a mind where there isn’t one.
No. 038PhenomenonArtificial Intelligence
The 1960s chatbot people confided in ELIZA just reflected your words back as questions. Its creator’s secretary still asked him to leave the room so she could talk to it in private.
IX
Then the promises outran the results, and the money ran out, twice.
No. 003ConceptArtificial Intelligence
AI has gone through “winters” before In 1970 a founder of AI predicted human-level machines within three to eight years. Within a few years funding collapsed, and it collapsed again in the late 1980s.
X
Finally, data, gaming chips and patience brought neural networks back, all the way to today’s AI assistants.
No. 033EventArtificial Intelligence
From a bedroom GPU to ChatGPT In 2012 a neural network trained on two gaming graphics cards in a student’s bedroom crushed an image-recognition contest. Ten years later, chatbots built on the same lineage reached the public.