The Origin of Artificial Intelligence: from the idea of thinking machines to the technological revolution.
Updated: 3 hours ago
Discover the origins of Artificial Intelligence, from Alan Turing's ideas to the Dartmouth conference in 1956, when the field gained its name and identity.
Introduction
To talk about the origin of Artificial Intelligence is to talk about an age-old question: can a machine think? Long before virtual assistants, self-driving cars, or tools capable of generating text and images existed, mathematicians, philosophers, and engineers were already trying to understand whether reasoning could be described by rules and reproduced by an artifact.
AI wasn't born from a single invention. It emerged from the intersection of several fields: philosophy, mathematics, logic, psychology, linguistics, and computer science. Therefore, its history begins before the modern computer and spans decades of experiments, expectations, and shifts in direction.
Although there are much older precedents, the official birth of Artificial Intelligence as a field of research is usually associated with 1956, when a group of scientists met at Dartmouth College in the United States. It was on this occasion that the expression "artificial intelligence" gained prominence and came to identify a new scientific program.
The idea of artificial intelligence before computers.
The desire to create artificial beings capable of acting intelligently appears in different cultures and historical periods. Mechanical automatons, articulated dolls, and calculating machines sparked the imagination because they seemed to imitate human actions. However, these devices typically repeated predefined movements. They did not learn or solve problems in general.
The great transformation occurred when thinkers began to treat reasoning as something that could be represented by symbols and rules. If a person can follow a logical procedure to solve a problem, perhaps a machine can do it too, provided it receives the right instructions.
This view was strengthened by the development of mathematical logic between the 19th and 20th centuries. Researchers began to investigate how statements could be formally represented and how new conclusions could be drawn from rules. This work helped build the conceptual foundation of computing and, later, of early AI programs.
Still, it's important to separate two things: the existence of ideas about intelligent machines and the creation of Artificial Intelligence as a discipline. For a long time, there were philosophical speculations and mechanical devices, but there were no programmable computers flexible enough to test these hypotheses.
Alan Turing and the question "Can machines think?"
One of the most important names in this transition was the British mathematician Alan Turing. During World War II, Turing participated in fundamental work for computing and codebreaking. After the war, he also dedicated himself to questions about how computers work and their possibilities.
In 1950, Turing published the article “Computing Machinery and Intelligence” in the philosophical journal Mind. The text began with a direct question: “Can machines think?”. Turing realized, however, that it would be difficult to arrive at a consensual definition of “thinking”. Instead of trying to resolve this question using only abstract concepts, he proposed a practical way to investigate it.
His proposal later became known as the Turing Test. In the experiment, a person converses in writing with two participants without knowing which one is human and which one is a machine. If the evaluator cannot consistently distinguish the two based on the responses, the machine would demonstrate behavior indistinguishable from a human in that context.
Turing's goal was not to assert that a machine would have consciousness or feelings. His proposal shifted the focus of the discussion: instead of asking what happens "inside" the machine, it focused on what it was capable of doing in a conversation. This approach profoundly influenced discussions about artificial intelligence and language.
Turing also suggested that computers could learn. He envisioned a kind of "child machine," which would start with limited capabilities and be improved through training. This idea is surprisingly close to certain principles later used in machine learning, although current methods rely on very different data, statistics, and computational resources.
The computer as a universal machine
To understand why Turing's ideas were so important, it's necessary to remember that the first computers weren't seen as tools for conversing, creating images, or recognizing objects. They were enormous, expensive machines specialized in calculations.
Turing helped solidify a broader vision: a machine could perform different tasks if it had memory to store information and instructions. Instead of building a new machine for each problem, it would be possible to program the same machine to perform calculations, manipulate symbols, play games, or process languages.
This flexibility is essential for AI. A traditional calculator performs very specific operations. A programmable computer, on the other hand, can represent different types of problems and follow varied strategies. Artificial intelligence depends precisely on this combination of representation, memory, rules, and processing capacity.
At the time, however, the hardware was still limited. Memory was small, computers occupied entire rooms, and programs needed to be written very carefully. Therefore, early AI research focused on well-defined problems, such as games, logic proofs, and theorem solving.
The Dartmouth conference and the official birth of AI.
The most cited milestone in the origins of Artificial Intelligence occurred in 1956 at Dartmouth College in Hanover, New Hampshire. The meeting was organized by John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon. The proposal had been submitted to the Rockefeller Foundation the previous year and is considered one of the first documented appearances of the term "artificial intelligence".
The researchers defended an ambitious hypothesis: in principle, all aspects of learning and intelligence could be described with sufficient precision for a machine to simulate them. The project mentioned tasks such as using language, forming concepts, solving problems previously reserved for humans, and improving one's own performance.
The meeting did not produce a ready-made intelligent machine, nor a single system capable of fulfilling all those promises. Its significance was different. It brought together researchers interested in treating intelligent behavior as an engineering and scientific investigation problem. From then on, "artificial intelligence" began to function as the name of an organized field.
The conference also helped bring together researchers working with different approaches. Some argued that intelligence could be reproduced through symbols and logical rules. Others were interested in learning, pattern recognition, and brain-inspired models. This division between symbolic and connectionist approaches would accompany the history of AI for many decades.
The first Artificial Intelligence programs
One of the first notable results of this period was Logic Theorist, developed by Allen Newell, Herbert Simon, and Cliff Shaw. The program was able to demonstrate some theorems of propositional logic. Although limited, it showed that a computer could manipulate symbols and find solutions by following a strategy, rather than simply performing arithmetic calculations.
Another important field was that of games. Chess became a laboratory for AI because it involves planning, prediction, and choosing between alternatives. Researchers imagined that if a machine could evaluate positions and select good moves, it might be possible to better understand reasoning in general.
There was also work on problem-solving, machine translation, and language comprehension. Many of these projects received funding because they seemed to promise rapid advances. There was an atmosphere of great confidence: some researchers believed that a general intelligence similar to that of humans could appear within a few decades.
Over time, it became clear that tasks that seemed simple for a person were very difficult for a computer. Recognizing an object in a photograph, interpreting an ambiguous sentence, or understanding the context of a conversation requires an enormous amount of knowledge and implicit decisions. This contrast between human ease and computational difficulty became known as one of the great challenges of AI.
The importance of neural networks and learning.
Not all early research followed the line of symbolic logic. Other scientists attempted to create models closer to the functioning of the brain, using connected units that received signals and adjusted their responses.
These models gave rise to artificial neural networks. Such a system does not need to receive an explicit rule for each case. It can adjust its parameters based on examples. This process is related to machine learning, an area that would later become central to the development of AI.
Early neural networks were quite limited by the available computing power and the amount of data. For this reason, during some periods they lost ground to rule-based systems. Decades later, with faster processors, large databases, and new mathematical methods, neural networks have regained momentum.
The history of AI, therefore, is not a straight line. It alternates between periods of enthusiasm and moments of frustration. When promises were not fulfilled, funding decreased and research entered phases known as "AI winters." When new methods or greater processing power emerged, interest returned.
Why does the origin of AI remain important?
Understanding the origins of Artificial Intelligence helps us comprehend what the technology is—and also what it is not. AI didn't suddenly appear with today's applications. It's the result of a long attempt to transform aspects of reasoning into representations that a machine can process.
History also shows that artificial intelligence does not necessarily mean consciousness, emotions, or human understanding. Early systems were capable of performing specific tasks using symbols, rules, and strategies. Even when an AI provides convincing answers, this does not prove that it thinks in the same way as a person.
Another important lesson is that the advancement of AI depends on more than just a good idea. It requires mathematical models, data, computers, energy, programming, and careful evaluation. Progress occurs when these elements combine.
The Dartmouth conference gave the field its name and identity, but it did not end the discussion about the meaning of intelligence. On the contrary: since then, each new technology reopens Turing's original question. If a machine can converse, recognize patterns, play games, translate, or create content, which of these abilities should be called intelligence?
Conclusion
The origin of Artificial Intelligence lies in a combination of ancient ideas and modern technologies. Philosophy raised the possibility of machines imitating human behavior. Logic offered ways to represent rules. Mathematics described formal processes. Computers made it possible to execute these instructions on a large scale.
Alan Turing was instrumental in transforming the question "can machines think?" into a problem that could be investigated through behavior. In 1956, the Dartmouth conference brought together researchers and consolidated the name of a new scientific field. The first programs, neural networks, and experiments with games established paths that are still present in current tools.
Contemporary AI is far more powerful than its pioneers could have imagined, but it still carries the fundamental questions of its origin. Understanding this past allows us to look at the present with greater clarity and realize that each technological advance is also part of a larger story: humanity's attempt to understand intelligence and build machines capable of reproducing some of its aspects.
Suggestions for image captions
Original conceptual illustration produced for this article. The image represents the encounter between early computers, mathematical logic, and the desire to understand human intelligence.
Alan Turing in 1951. Photograph by Elliott & Fry, available in the public domain via Wikimedia Commons.
Alan Turing at age 16. Photograph by an unknown author, available in the public domain via Wikimedia Commons.
Image credits
Conceptual illustration: original image generated for this article, 2026.
Alan Turing in 1951: Elliott & Fry, photograph available on Wikimedia Commons, identified as public domain.
Alan Turing at age 16: author unknown, photograph available on Wikimedia Commons, identified as public domain.
References
Comments