history · 76 years in 13 moments

How we got here.

From the Turing test to continuous learning. Every breakthrough, every winter, every paper that changed what's possible.

Explore timeline → Deep dives

eras foundationswinterrebirthdeep erascaleagentsnow

1950 foundations

Turing test

Alan Turing reframes 'can machines think?' into 'can they convince you?'

↗ deep dive

1956 foundations

Dartmouth workshop

John McCarthy coins 'artificial intelligence.' The field begins.

↗ deep dive

1957 foundations

Perceptron

Frank Rosenblatt builds the first trainable neural network.

↗ deep dive

1974 winter

First AI winter

Funding cuts after Minsky/Papert's Perceptrons paper.

↗ deep dive

1986 rebirth

Backprop popularized

Rumelhart, Hinton, Williams. Deep nets become trainable.

↗ deep dive

1997 rebirth

Deep Blue beats Kasparov

Chess falls to brute search + handcrafted heuristics.

↗ deep dive

2012 deep

AlexNet · ImageNet

Deep learning sweeps vision. The era begins.

↗ deep dive

2017 deep

'Attention is all you need' ★

Transformer paper. Everything afterwards rests here.

↗ deep dive

2020 scale

GPT-3

175B params. Few-shot learning becomes practical.

↗ deep dive

2022 scale

ChatGPT ★

Public discovers LLMs. 100M users in 60 days.

↗ deep dive

2023 agents

Multimodal arrives

GPT-4V, Gemini, Claude 3 vision. Tools, agents.

↗ deep dive

2025 agents

Frontier consolidation

Four labs at the top; open weights catch up to within 6 months.

↗ deep dive

2026 now

Continuous learning ★

Three labs converge on post-pretraining loops within 72h of each other.

↗ deep dive

by artifact

Dig deeper.

▲

Breakthroughs

Moments that mattered

start here

The history of AI in 90 minutes.

8 lessons. From perceptron to transformer. No prerequisites. Understand where we came from to see where we're going.

Begin →