Skip to content
SI.info

Reference

History of Super Intelligence: A Timeline from 1943 to Today

70 moments that took the field from the first mathematical model of a neuron in 1943 to frontier systems, and to the day Washington renamed it. Every entry links to a deeper explainer where we have one.

Published
ResearchModelsPolicyIndustryCulture

1943–1979: The founding era

  1. Research

    The first mathematical model of a neuron

    Warren McCulloch and Walter Pitts show how networks of simple "neurons" could compute logical functions, the seed of neural networks.

  2. Research

    Turing asks "Can machines think?"

    Alan Turing's paper "Computing Machinery and Intelligence" proposes the imitation game, later called the Turing test.

  3. Research

    The Dartmouth workshop names "artificial intelligence"

    John McCarthy, Marvin Minsky, Nathaniel Rochester and Claude Shannon convene a summer workshop at Dartmouth College under the name "artificial intelligence", coined in their 1955 proposal.

  4. Research

    Rosenblatt's perceptron

    Frank Rosenblatt demonstrates the perceptron, an early machine that learns to classify inputs.

  5. Research

    "Machine learning" enters the vocabulary

    IBM's Arthur Samuel popularises the term with a checkers program that improves by playing against itself.

  6. Research

    I. J. Good predicts an "intelligence explosion"

    The statistician argues that an "ultraintelligent machine" could design better machines, leaving human intelligence far behind.

  7. Culture

    ELIZA, the first famous chatbot

    Joseph Weizenbaum's ELIZA imitates a psychotherapist, and surprises its creator by how readily people confide in it.

  8. Industry

    The first "winter"

    After early hype outpaces results, funding dries up in the U.S. and U.K., beginning the first period of reduced interest in the field.

1980–2009: Expert systems to statistics

  1. Industry

    The expert-systems boom

    Rule-based "expert systems" spread through business. Their brittleness leads to a second downturn later in the decade.

  2. Research

    Backpropagation goes mainstream

    Rumelhart, Hinton and Williams publish an influential paper on training multi-layer neural networks with backpropagation.

  3. Culture

    Vinge's "Coming Technological Singularity"

    Mathematician and novelist Vernor Vinge predicts superhuman intelligence within 30 years and popularises the idea of a singularity.

  4. Research

    Deep Blue beats Garry Kasparov

    IBM's chess computer defeats the reigning world champion in a six-game match, a landmark for machine intelligence.

  5. Industry

    Nvidia introduces CUDA

    CUDA makes graphics chips programmable for general computing, laying the groundwork for GPU-powered deep learning.

2010–2016: The deep learning breakthrough

  1. Industry

    DeepMind is founded in London

    Demis Hassabis, Shane Legg and Mustafa Suleyman start DeepMind with the goal of solving intelligence.

  2. Culture

    IBM Watson wins Jeopardy!

    Watson defeats champions Ken Jennings and Brad Rutter on the U.S. quiz show.

  3. Research

    AlexNet ignites the deep-learning era

    A neural network by Alex Krizhevsky, Ilya Sutskever and Geoffrey Hinton wins the ImageNet challenge by a wide margin, trained on GPUs.

  4. Industry

    Google acquires DeepMind

    The acquisition brings one of the leading research labs inside Google.

  5. Culture

    Bostrom's "Superintelligence" is published

    Nick Bostrom's book defines superintelligence and brings the control problem to a wide audience.

  6. Industry

    OpenAI is founded

    A non-profit research lab launches with the mission of ensuring artificial general intelligence benefits all of humanity.

  7. Research

    AlphaGo defeats Lee Sedol

    DeepMind's system wins 4–1 against one of the world's best Go players, years earlier than experts expected.

2017–2021: Transformers and scale

  1. Research

    "Attention Is All You Need"

    Google researchers introduce the transformer, the architecture behind virtually every modern language model.

  2. Models

    GPT-1

    OpenAI shows that a transformer pre-trained on lots of text can be adapted to many language tasks.

  3. Models

    GPT-2 and the staged release

    OpenAI initially withholds its full 1.5-billion-parameter model over misuse concerns, sparking a debate about openness.

  4. Models

    GPT-3

    At 175 billion parameters, GPT-3 shows that scaling up produces surprising new abilities, such as learning tasks from a few examples.

  5. Research

    AlphaFold 2 solves protein folding

    DeepMind's system predicts protein structures with near-experimental accuracy at the CASP14 competition.

  6. Industry

    Anthropic is founded

    Former OpenAI researchers led by Dario and Daniela Amodei start a safety-focused lab.

2022–2025: The generative boom

  1. Models

    DALL·E 2 brings text-to-image to the mainstream

    Photo-realistic images from text prompts capture public attention, followed by Midjourney and Stable Diffusion.

  2. Models

    ChatGPT launches

    OpenAI's chatbot reaches an estimated 100 million users in about two months and starts the generative boom.

  3. Models

    GPT-4

    OpenAI's multimodal model performs at human level on many professional exams.

  4. Culture

    The "pause" open letter

    Thousands, including Elon Musk and Yoshua Bengio, call for a six-month pause on training systems more powerful than GPT-4.

  5. Culture

    Leaders warn of extinction risk

    Hundreds of researchers and executives sign a one-sentence statement: mitigating the risk of extinction from the technology should be a global priority.

  6. Policy

    First White House voluntary commitments

    Seven companies, including OpenAI, Google, Anthropic and Meta, make voluntary safety pledges to the Biden administration.

  7. Policy

    Executive Order 14110

    President Biden signs a sweeping order on safety, including reporting requirements for models trained above 10^26 operations.

  8. Policy

    Bletchley Park safety summit

    The U.K. hosts the first global summit on frontier-model safety; 28 countries and the EU sign the Bletchley Declaration.

  9. Industry

    OpenAI's board fires Sam Altman, then reinstates him

    A five-day crisis exposes tensions between safety governance and commercial growth at the leading lab.

  10. Industry

    Safe Superintelligence Inc. is founded

    Ilya Sutskever launches a company with "one goal and one product": safe superintelligence.

  11. Policy

    The EU AI Act enters into force

    The world's first comprehensive law on the technology begins its phased implementation.

  12. Models

    OpenAI previews o1, the first major reasoning model

    Models that "think" before answering open a new axis of progress, spending compute at answer time.

  13. Research

    Nobel Prizes for neural-network pioneers

    Geoffrey Hinton and John Hopfield win the Physics prize; Demis Hassabis and John Jumper share Chemistry for AlphaFold.

  14. Models

    DeepSeek-R1 shocks markets

    A low-cost open-weights reasoning model from China rivals the best U.S. systems; Nvidia's market value drops sharply days later.

  15. Policy

    The 2023 safety order is revoked

    On his first day back in office, President Trump revokes Executive Order 14110.

  16. Industry

    Stargate is announced

    OpenAI, SoftBank and Oracle announce a giant U.S. infrastructure venture at the White House.

  17. Industry

    Mira Murati founds Thinking Machines Lab

    OpenAI's former CTO launches a new lab with a team of top researchers.

  18. Policy

    EU bans on unacceptable-risk systems apply

    Social scoring and other prohibited practices become illegal across the EU.

  19. Models

    Claude 4

    Anthropic's Claude Opus 4 and Sonnet 4 push coding and agentic capabilities forward.

  20. Industry

    Meta Superintelligence Labs is formed

    Mark Zuckerberg reorganises Meta's efforts under Alexandr Wang with the explicit goal of superintelligence.

  21. Policy

    America's AI Action Plan

    The White House publishes more than 90 federal actions to accelerate innovation and build infrastructure.

  22. Models

    GPT-5

    OpenAI unifies its fast and reasoning models into a single flagship.

  23. Policy

    California signs SB 53

    The first U.S. state law aimed at frontier models requires safety frameworks and incident reporting.

  24. Culture

    Statement on Superintelligence

    Tens of thousands of signatories, including Geoffrey Hinton and Yoshua Bengio, call for a prohibition on developing superintelligence until it can be done safely.

  25. Industry

    OpenAI becomes a public benefit corporation

    A restructuring creates OpenAI Group PBC under the OpenAI Foundation, with Microsoft holding about 27%.

  26. Industry

    Microsoft forms a superintelligence team

    Mustafa Suleyman announces the MAI Superintelligence Team focused on "humanist superintelligence".

  27. Policy

    The Genesis Mission launches

    Executive Order 14363 directs the Department of Energy to pair national-lab supercomputers with scientific data and SI.

  28. Policy

    Executive Order 14365 targets state laws

    The White House sets a policy of minimally burdensome national rules and moves against conflicting state laws.

  29. Policy

    New York signs the RAISE Act

    New York becomes the second state to regulate frontier models, effective 2027.

2026–today: The SI era

  1. Industry

    SpaceX acquires xAI

    An all-stock deal values xAI at about $250 billion; it later becomes SpaceXAI.

  2. Policy

    National Policy Framework sent to Congress

    The White House recommends a single federal law that preempts state rules on SI development.

  3. Models

    Meta launches Muse Spark

    Meta Superintelligence Labs' first flagship model marks Meta's shift from open Llama to closed models.

  4. Models

    DeepSeek V4 weights released

    DeepSeek publishes MIT-licensed preview weights of its trillion-parameter-class V4 models.

  5. Policy

    Colorado repeals and replaces its AI Act

    The first broad state law on high-risk systems is swapped for a narrower version before it ever takes effect.

  6. Policy

    EU delays high-risk rules

    The AI Omnibus amendment enters into force, pushing most high-risk obligations to December 2027 and August 2028.

  7. Models

    The largest open-weights model yet

    Alibaba releases the weights of its 2.4-trillion-parameter Qwen3.8-Max.

  8. Models

    Claude Fable 5.1

    Anthropic launches its most capable model for demanding reasoning and long-horizon agents.

  9. Models

    GPT-6 Astra

    OpenAI's flagship launches, first to vetted cyber-defenders, as the first model rated "Critical" for cyber capability under its framework.

  10. Policy

    "Hereinafter officially called Super Intelligence"

    At the UN General Assembly, President Trump announces that AI will be renamed Super Intelligence.

  11. Policy

    A bill to ban superintelligence

    Sen. Bernie Sanders and Rep. Greg Casar propose banning artificial superintelligence and creating a Department of AI.

  12. Policy

    White House Accord on Super Intelligence

    Leaders of Google, Anthropic, Meta, OpenAI, xAI and Nvidia sign a voluntary pledge of four-layer oversight.

  13. Policy

    Executive Order 14434 renames AI "Super Intelligence"

    Federal agencies are ordered to use "Super Intelligence" and "SI" in place of "Artificial Intelligence" and "AI". The SI era officially begins.

  14. Models

    Google announces Gemini 4 Argon

    Google's next-generation model goes first to trusted testers and cyber-defenders.

  15. Policy

    EO 14434 published in the Federal Register

    The order appears as FR Doc. 2026-20321. The 60-day clock for a federal definition of SI runs to about November 28.

Free newsletter

Get the next chapter by email

One short email a week: what changed in Super Intelligence, policy and models — and why it matters.

Free. One email a week. Sent via beehiiv, which counts opens and clicks. Unsubscribe anytime. Privacy policy