How did we get from a thought experiment to governments and trillion-dollar companies openly racing toward super intelligence? This timeline tracks the key moments, from Alan Turing to the September 2026 decision to rename AI “SI”. It ends with what experts predict comes next. We update it as new events happen.
Part 1: The ideas (1950–1999)
| Year | Milestone |
|---|---|
| 1950 | Alan Turing publishes “Computing Machinery and Intelligence” and proposes the imitation game (the Turing test). |
| 1956 | The Dartmouth workshop coins the term “artificial intelligence”. AI becomes a field. |
| 1965 | I. J. Good describes an “ultraintelligent machine” and an “intelligence explosion”, the first clear description of super intelligence. |
| 1970s–80s | Two “AI winters”: funding collapses after early promises go unmet. |
| 1993 | Vernor Vinge predicts superhuman intelligence and the technological singularity “within thirty years”. |
| 1997 | IBM’s Deep Blue beats world chess champion Garry Kasparov. It’s the first famous superhuman narrow AI. |
| 1998 | Nick Bostrom defines superintelligence in “How Long Before Superintelligence?” |
Part 2: The deep learning era (2000–2021)
| Year | Milestone |
|---|---|
| 2005 | Ray Kurzweil’s The Singularity Is Near predicts human-level AI by 2029 and the singularity by 2045. |
| 2012 | AlexNet wins the ImageNet contest by a wide margin, and the deep learning boom begins. |
| 2014 | Bostrom’s Superintelligence: Paths, Dangers, Strategies becomes a bestseller. Elon Musk, Bill Gates and Stephen Hawking warn about AI risk. |
| 2015 | OpenAI is founded as a nonprofit with the mission of making AGI benefit humanity. |
| 2016 | DeepMind’s AlphaGo beats Lee Sedol at Go, a decade earlier than many expected. |
| 2017 | Google researchers publish “Attention Is All You Need”, introducing the Transformer that powers modern chatbots. |
| 2020 | GPT-3 shows that scaling language models unlocks broad abilities. AlphaFold solves protein structure prediction. |
Part 3: The race begins (2022–2025)
| Date | Milestone |
|---|---|
| Nov 2022 | ChatGPT launches and reaches 100 million users in about two months. |
| Mar 2023 | GPT-4 is released. The Future of Life Institute’s open letter calls for a six-month pause. |
| May 2023 | Hundreds of AI leaders sign a statement that AI extinction risk should be a “global priority”. |
| Jul 2023 | OpenAI launches a “Superalignment” team for controlling AI smarter than humans. |
| Jun 2024 | Ilya Sutskever founds Safe Superintelligence Inc. (SSI). |
| Oct 2024 | AlphaFold’s creators share the Nobel Prize in Chemistry. |
| Apr 2025 | The AI 2027 scenario predicts superintelligence by the end of 2027. |
| Jun 2025 | Meta creates Meta Superintelligence Labs. |
| Oct 22, 2025 | The Statement on Superintelligence calls for a ban until it is proven safe. |
Part 4: The year of “SI” (2026)
| Date | Milestone |
|---|---|
| Jan 2026 | At Davos, Elon Musk says AI could be smarter than any single human by the end of 2026 or in 2027. |
| Feb 2026 | SpaceX acquires xAI. |
| Sept 22, 2026 | At the UN, President Trump says AI will be called “super intelligence” in US documents. |
| Sept 29, 2026 | Trump signs the order Inaugurating The Era Of Super Intelligence. Federal agencies switch from “AI” to “SI”. (Full story) |
| Oct 4, 2026 | Musk posts “No more AI… SI” and says xAI’s successor will become SpaceXSI. (Full story) |
Note that the 2026 changes are about naming. No system has been shown to meet the technical definition of super intelligence. See super intelligence vs AI.
Part 5: What comes next (predictions)
| Prediction | Who | Year |
|---|---|---|
| AI smarter than any single human | Elon Musk (Davos 2026) | 2026–2027 |
| Superintelligence, fast scenario | AI 2027 (Kokotajlo et al.) | ~2027–2028 |
| Human-level AI (Turing test passed) | Ray Kurzweil | 2029 |
| AI smarter than all humans combined | Elon Musk | ~2030–2031 |
| The singularity | Ray Kurzweil | 2045 |
| Human-level AI (median of researcher surveys) | Academic surveys | Roughly 2040s–2060s |
| Not with current techniques | Many skeptics | Uncertain |
Why the predictions differ so much
- Scaling optimists see steady gains from more compute and data and expect them to continue.
- Takeoff believers think that once AI can do AI research, progress compounds (see AGI vs ASI).
- Skeptics point to unreliability, limits on data and energy, and the gap between passing tests and doing real-world work.
One clear trend: forecasts keep getting earlier. Many experts who said “50+ years” in 2015 now say “10–20 years” or less.
How to read this timeline
- Milestones come in bursts. Long quiet periods (AI winters) were followed by sudden leaps (2012, 2017, 2022).
- Narrow superhuman AI has existed for decades. General super intelligence hasn’t yet.
- Language changed before technology did. In 2026, “super intelligence” became an official name before it became a reality.
We’ll keep adding events as they happen. Bookmark this page, or follow our super intelligence news hub.
Enjoyed this? Get the SI Briefing
One short email a week: what changed on the road to super intelligence, and why it matters. No spam, unsubscribe anytime.
Frequently asked questions
How long until superintelligence?
Nobody knows. Some AI leaders and the AI 2027 scenario suggest before 2030; many academic surveys put human-level AI around mid-century, with superintelligence later. Forecasts have moved earlier in recent years.
How far away are we from super intelligence?
No current AI is superintelligent. Today’s systems are strong but unreliable generalists. The remaining distance depends on whether AI can soon speed up its own research, which is the main point of disagreement.
When was the idea of superintelligence first proposed?
In 1965, mathematician I. J. Good described an “ultraintelligent machine” that could trigger an “intelligence explosion”. Nick Bostrom formalized the term superintelligence in 1998.
Will humans be immortal by 2030?
Very unlikely. Some futurists such as Ray Kurzweil predict dramatic life extension through AI and biotech, but there is no scientific evidence that human immortality will be achieved by 2030.
What happened in AI in 2026?
Among other things, SpaceX absorbed xAI, the US government renamed AI “Super Intelligence (SI)” in September 2026, and Elon Musk announced xAI’s successor would be called SpaceXSI.