gensai.aiSince ChatGPT arrived in late 2022, artificial intelligence seems to have become omnipresent. Many people discovered AI at that moment and sometimes have the impression that it is a recent technology.
In reality, the concept of artificial intelligence has existed for nearly 70 years. The term itself was coined in 1956, at an event that became historic in the research world: the Dartmouth conference, in the United States.
Understanding where the term comes from helps make better sense of what AI actually is today — and why its development is only accelerating now.
When was the term "artificial intelligence" created?
The term "artificial intelligence" officially appeared in 1956. It was proposed by the American researcher John McCarthy, who that year organised a scientific conference at Dartmouth College, in New Hampshire.
The aim of the gathering was ambitious: bringing several researchers together to explore the possibility of getting machines to perform tasks that normally require human intelligence.
In the research proposal accompanying the conference, the organisers set out a bold hypothesis:
"Every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it."
At the time, computing was still in its infancy. Computers took up entire rooms and had extremely limited processing power. Yet this conference laid the foundations of a new scientific field: artificial intelligence.
Who were the pioneers of artificial intelligence?
The Dartmouth conference brought together several researchers who would become major figures in the field.
Among them:
- John McCarthy, often considered the father of the term "artificial intelligence"
- Marvin Minsky, a cognitive science specialist and co-founder of the MIT AI laboratory
- Herbert Simon and Allen Newell, researchers who would develop some of the first programs capable of reasoning
These researchers were already imagining machines able to:
- play chess
- solve logical problems
- understand language
- learn from data
Some of these ambitions would only materialise several decades later.
Did artificial intelligence exist before 1956?
Yes — ideas around intelligent machines are far older than the term itself.
As early as the 1940s and 1950s, several researchers were already working on concepts close to AI:
- Alan Turing, the British mathematician, conceived the famous Turing test as early as 1950, designed to determine whether a machine could imitate human conversation.
- The first artificial neural networks were theorised in 1943 by Warren McCulloch and Walter Pitts.
- The first programmable computers appeared after the Second World War.
The term "artificial intelligence" therefore mainly served to give structure to a research field that already existed in various forms.
Why didn't AI take off earlier?
Despite the early enthusiasm, progress in AI was limited for a long time. Several periods known as "AI winters" even slowed research down.
The main reason is simple: the technology was not ready.
Three key elements were missing:
1. Computing power
The computers of the 1960s and 1970s were extremely limited.
Today, processors, GPUs and cloud infrastructure make it possible to train models with billions of parameters.
2. Data
Modern AI models need enormous volumes of data to learn.
With the rise of the internet, smartphones and digital services, that data is now available at scale.
3. Algorithmic advances
Certain key techniques, such as deep learning, only really started producing spectacular results from the 2010s onwards.
These three factors — computing power, data and new algorithms — explain why AI is progressing so quickly today.
Why does artificial intelligence seem to be appearing only now?
Artificial intelligence is therefore an old field, but its mass adoption is recent.
The real turning point came in the 2010s with several major advances:
- spectacular progress in image recognition
- improvements in voice assistants
- the development of deep learning
Then, in 2022, the arrival of ChatGPT democratised access to generative AI for the general public.
For the first time, millions of users could converse directly with an artificial intelligence model, generating text, code or images.
This accessibility explains why so many people discovered AI at that point, even though the discipline has existed for several decades.
What does "artificial intelligence" really mean today?
The term "artificial intelligence" can sometimes be confusing.
In most cases, today's systems do not genuinely "think" like humans. They rest instead on statistical and probabilistic models trained on very large quantities of data.
For example, the LLMs (Large Language Models) used in conversational agents work by predicting the probability of the next word in a sentence.
In other words, these systems have no conscious understanding of the world. They learn structures in data and generate answers based on those probabilities.
This characteristic explains well-known phenomena such as AI hallucinations, where the model produces a plausible but incorrect answer.
Why does the history of AI matter for businesses?
The history of artificial intelligence shows that this technology is not a recent fad, but the result of several decades of scientific research.
It also serves as a reminder of an essential point: AI systems are not infallible.
Because they rest on probabilistic models, they always carry a degree of uncertainty. This is why enterprise AI projects generally need to include:
- human supervision
- reliable knowledge bases (RAG)
- verification mechanisms
Used well, artificial intelligence can automate certain tasks, make information easier to reach or improve the user experience. But it must be deployed with method, and with an understanding of its limits.