When was AI invented? Most pages stop at 1956, but the name "artificial intelligence" is a year older than the field it describes. I went back to the original 1955 proposal, the first working program and the two crashes that nearly ended AI, to answer one question: how did a summer workshop turn into the chatbot on your phone?
⚡ Short answer: AI was invented as a research field in summer 1956, at the Dartmouth Summer Research Project on Artificial Intelligence. The name itself first appeared in the workshop proposal dated August 31, 1955, written by John McCarthy, Marvin Minsky, Nathaniel Rochester and Claude Shannon. The Logic Theorist, widely called the first AI program, was presented there.
Some links are affiliate links · they support this site at no extra cost to you · Full disclosure
When Was AI Invented? The Short Answer
AI was invented in 1956, when a small group of researchers spent the summer at Dartmouth College in Hanover, New Hampshire, testing one bold idea: that every part of learning could be described precisely enough for a machine to copy it. That workshop is the moment AI stopped being a thought experiment and became a field with a name, a goal and its first program.
So why do some sources say 1950, or even 1943? Because "when was AI invented" has three honest answers, depending on whether you mean the idea, the name or the working software.
What year was AI invented?
1956, at the Dartmouth summer workshop. If you mean the word, 1955.Where was AI invented?
Dartmouth College in Hanover, New Hampshire, in the United States.Why was AI invented?
To test whether every feature of learning could be described precisely enough for a machine to simulate it.Was AI invented in 1950?
No. 1950 is Turing's famous paper. The field and its name came five to six years later.Who was AI invented by?
A team, not one person: McCarthy, Minsky, Rochester and Shannon organized it, Newell and Simon brought the first program.Was AI invented before computers?
The ideas were, the field wasn't. The 1943 neuron model came before most electronic computers existed.Who Invented AI?
No single person invented AI. John McCarthy is the closest answer, because he picked the name and organized the workshop. In his own account, McCarthy explains that he, Minsky, Shannon and Rochester (the IBM engineer who designed the IBM 701) wrote the funding request to the Rockefeller Foundation in August 1955, and that this proposal is where the term "artificial intelligence" comes from. You can read his recollection on his Stanford page.
The honest credit list is longer. These six names are the ones I would put on the poster:
If someone asks you for "the father of AI", McCarthy is the standard answer and Turing is the defensible alternative. Both are right about different things: Turing framed the question, McCarthy turned it into a discipline. Britannica's overview of Turing's early work shows how much of the agenda he sketched before the field had a name.
AI Timeline: When Was AI Invented, and What Came Next?
Counting from Dartmouth, AI is 70 years old. Here is the whole story in 12 dates. Filter by era to see how the field rose, crashed twice and came back on a completely different idea.
One footnote worth knowing: in 2024 the Nobel Prize in Physics went to John Hopfield and Geoffrey Hinton for the neural-network work of the 1980s that today's machine learning is built on. The ideas that survived both winters ended up with the most famous prize in science. For a compact list of these milestones with photos, Live Science keeps a good one.
What Was the First AI Program?
If you ask "when was AI invented" as working software rather than as an idea, the answer is a program. The Logic Theorist, built by Allen Newell, Herbert Simon and Cliff Shaw at the RAND Corporation in 1955–56, is the program most historians call the first AI program. It proved theorems from Whitehead and Russell's Principia Mathematica, eventually 38 of the first 52 in chapter two, and found a more elegant proof for one of them than Whitehead and Russell's own.
My favorite detail: before it ran on a computer, the team tested it by hand. Simon handed index cards to his wife, his children and some graduate students, and each person played one part of the program. The full story, including the lukewarm reaction it got at Dartmouth, is in the Logic Theorist history.
Why Did AI Almost Disappear? The Two AI Winters
The Dartmouth group expected big progress within a generation. Instead the early programs only worked inside tidy, toy-sized problems. Real language and real images exploded into too many possibilities, a problem researchers called combinatorial explosion, and the money followed the disappointment out the door.
Both winters had the same cause: hand-written rules can't cover the real world. The AI winter record lists the smaller crashes too, from failed machine translation in 1966 to the end of Japan's Fifth Generation project. IBM has a readable history of AI and its winters if you want the business side.
How Did Modern AI Finally Take Off?
AI didn't come back because someone finally wrote better rules. It came back because researchers stopped writing rules and let systems learn patterns from huge amounts of data. Three things had to arrive together: enough digital data, cheap parallel computing (graphics cards built for games), and training methods that kept improving as models grew.
| Moment | How it worked | What changed |
|---|---|---|
| Deep Blue, 1997 | Special hardware searching ~200M positions/sec | Proved raw computing could beat a world champion at one task |
| AlexNet, 2012 | Neural network trained on GPUs, learned from labeled images | Learning beat hand-built features; deep learning took over |
| Transformer, 2017 | "Attention" lets a model weigh every word against every other word | Made huge language models practical to train |
| ChatGPT, 2022 | A Transformer model tuned to follow instructions in chat | AI became a consumer product overnight |
If you prefer watching to reading, this 13-minute CrashCourse episode tells the comeback story through chess, from 1950s chess programs to Deep Blue and the deep learning revolution. It pairs well with the table above.
Deep Blue is the useful contrast. The record of the 1997 match shows a machine that searched harder than any human, but it couldn't do anything except chess. Modern systems trade that narrow certainty for broad, pattern-based skill. If the difference between "learning patterns" and "generating content" is new to you, my plain-English guide to machine learning vs generative AI covers it with everyday examples.
What This History Means for the AI Tools You Use
Knowing when AI was invented is trivia. Knowing how it was rebuilt explains most of the strange things your chatbot does. Today's assistants are the pattern-learning branch that won after the rule-based branch failed twice, and they carry that DNA into every answer.
| ❌ What people assume | ✅ What the history shows |
|---|---|
| "ChatGPT was the first AI." | It arrived 66 years after Dartmouth and 56 years after ELIZA. It was the first one most people used. |
| "AI understands what it says." | Modern AI predicts likely words from patterns. That is why it can sound sure and still be wrong. |
| "AI progress only goes up." | Two winters wiped out funding when hype beat results. Hype cycles are part of the record. |
| "The same question gets the same answer." | Pattern-based models sample from probabilities, so answers can shift between tries. |
Three habits follow directly from that history. First, check facts that matter: confident tone is a feature of the prediction method, not proof. My guide to AI hallucinations shows the fixes that work, and why AI sounds confident when it's wrong explains the mechanism.
Second, expect variation and give clear context. If you have noticed the same prompt producing different results, the reasons are in why AI answers change. Third, discount the hype. Every era of AI has had predictions that missed by decades; Simon and Newell expected a computer chess champion within ten years of 1958, and it took until 1997.
Try the Tools This History Built
All three assistants below run on the Transformer design from 2017, and each has a free plan, so you can feel the difference between them in ten minutes. If you're deciding which one to pay for, I compare the plans in which AI subscription is worth it.
Quiz: How Well Do You Know When AI Was Invented?
Five questions, all answered somewhere above. Most people miss question one, because almost every page says 1956. Teachers: feel free to use it in class.
1. When was the term "artificial intelligence" first written down?
2. Which program is usually called the first AI program?
3. What triggered the first AI winter in the UK?
4. How did Deep Blue beat Kasparov in 1997?
5. Which 2017 paper introduced the design behind ChatGPT?
Frequently Asked Questions: When Was AI Invented
🧭 Go from AI history to AI you can use
The name is from 1955, the chatbot is from 2022. These four pick up the story where this timeline ends.