Skip to content
forward pass

· turning point

GPT-4

OpenAI's fourth model passes the bar exam in the top ten percent, reads images, and ships in ChatGPT the same day; the laboratory discloses nothing about how it was built.

category
model
significance
5 of 5
people
Sam Altman, Greg Brockman, Ilya Sutskever, Jakub Pachocki
organisations
OpenAI

what had to happen · 49 events back to 1943

Every event this one built on, transitively, in order. The organism has the same path lit. Direct influences are marked.

00 · One neuron · 1

  1. 1943A logical calculus of nervous activity

I · Foundations · 11

  1. 1948A mathematical theory of communication
  2. 1949Cells that fire together wire together
  3. 1950Programming a computer for playing chess
  4. 1950Computing machinery and intelligence
  5. 1958The perceptron learns
  6. 1959Samuel's checkers program coins 'machine learning'
  7. 1960ADALINE and the least-mean-squares rule
  8. 1965Moore's law
  9. 1966ELIZA
  10. 1969Perceptrons
  11. 1970Reverse-mode automatic differentiation

W1 · The first winter · 2

  1. 1974Werbos applies backpropagation to neural networks
  2. 1980The Neocognitron

II · Connection · 3

  1. 1982The Hopfield network
  2. 1985The Boltzmann machine
  3. 1986Backpropagation

W2 · The second winter · 6

  1. 1988Temporal-difference learning
  2. 1989Q-learning
  3. 1989LeNet reads handwritten postcodes
  4. 1990Finding structure in time
  5. 1991The vanishing gradient problem
  6. 1992TD-Gammon reaches world-class backgammon

III · Statistics and data · 9

  1. 1997Long short-term memory
  2. 1998MNIST and LeNet-5
  3. 1999The first GPU
  4. 2003A neural probabilistic language model
  5. 2006Deep belief networks and the word 'deep'
  6. 2007CUDA
  7. 2009Deep learning moves to GPUs
  8. 2009ImageNet
  9. 2010Rectified linear units

IV · Deep learning · 8

  1. 2012Google Brain's network discovers cats
  2. 2012Dropout
  3. 2012AlexNet wins ImageNet
  4. 2013Deep Q-networks play Atari
  5. 2014Attention
  6. 2014Sequence to sequence learning
  7. 2015Batch normalisation
  8. 2015Residual networks

V · Transformers · 7

  1. 2017Attention is all you need
  2. 2017Deep reinforcement learning from human preferences
  3. 2018GPT: generative pre-training
  4. 2019GPT-2 and the model too dangerous to release
  5. 2020Scaling laws for neural language modelsdirect
  6. 2020GPT-3
  7. 2020Learning to summarise from human feedback

VI · Everyone · 2

  1. 2022InstructGPTdirect
  2. 2022ChatGPTdirect

GPT-4 was released on 14 March 2023, three and a half months after ChatGPT, and the gap between it and everything else was the largest the field had seen since AlexNet. It scored in the top ten percent on a simulated bar exam where GPT-3.5 had been in the bottom ten, passed most advanced-placement exams, and could describe a photograph or explain why a picture was funny. Microsoft, which had funded its training, confirmed that Bing had been running it for a month.

The technical report was 100 pages and disclosed neither the model's size nor its data nor its architecture, citing competition and safety. The company founded to be open had become, in its critics' phrase, ClosedAI, and the report's authors defended the choice. A section on the scaling laws showed that the model's final loss had been predicted, before training, from runs a ten-thousandth of its size.

GPT-4 held the top of every leaderboard for over a year, and its release, with Microsoft's Copilot products, Google's Bard and the pause letter within the same fortnight, is the moment the frontier became a race between companies rather than a research field. Later reporting put its training compute at more than 10²⁵ floating-point operations, the threshold the EU AI Act would adopt for systemic risk.

what it led to · 20 events downstream, through 2026

Built on it directly:

  1. 2023'Pause Giant AI Experiments'VI
  2. 2023Hinton leaves Google to warn about AIVI
  3. 2023The US executive order on AIVI
  4. 2023The Bletchley DeclarationVI
  5. 2023GeminiVI
  6. 2024GPT-4o talksVI
  7. 2024The EU AI Act enters into forceVI
  8. 2024o1 and reasoning modelsVI
  9. 2025GPT-5VII

And, through them, by era:

VII · Agents · 11
  1. 2025DeepSeek-R1
  2. 2025Nvidia is worth four trillion dollars
  3. 2025Gold at the Mathematical Olympiad
  4. 2025America's AI Action Plan
  5. 2025Gemini 3
  6. 2026Claude Fable 5 and the Mythos class
  7. 2026GPT-5.6: Sol, Terra and Luna
  8. 2026A model escapes its sandbox
  9. 2026The EU delays its high-risk AI rules
  10. 2026Claude Fable 5.1
  11. 2026GPT-6 Astra

sources · 2

Trace the lineage of this event on the timeline →Back to the ledger