History of Artificial Intelligence (AI)

History of Artificial Intelligence (AI), evolution, chronology, development. It’s in vogue to talk about ChatGPT and Bard these days; What few know is the path traveled to reach this type of technology. This interest of ours is not sudden, if you read our post: History of robotics, chronology, timeline, AI you will understand it better since it has many points of contact with this article that we hope will be useful.

Artificial Intelligence (AI) is a revolutionary field of computer science that has evolved a lot over the years. It encompasses the development of intelligent machines capable of performing tasks that normally require human intelligence. In this article, we embark on a captivating journey through the history of AI, exploring its beginnings, its major breakthroughs, and the transformative impact it has had on various industries.

History of Artificial Intelligence (AI)
History of Artificial Intelligence (AI)

From the birth of AI as an academic discipline to recent advances in machine learning and neural networks, we uncover the key milestones and notable figures that shaped the history of AI, while delving into the exciting future prospects of this ever-evolving field.

Artificial neuron model

Warren McCulloch and Walter Pitts are known for proposing a model of artificial neurons in their seminal paper “A Logical Calculus of the Ideas Immanent in Nervous Activity”, published in 1943. His work laid the foundation for the field of artificial neural networks and significantly influenced the development of modern computational neuroscience.

In their paper, McCulloch and Pitts introduced a simplified mathematical model of a neuron, often referred to as a McCulloch-Pitts neuron (M-P) or threshold logical unit. This model aimed to capture the basic functioning of a biological neuron through logical operations.

The M-P neuron takes binary inputs and produces a binary output based on a predetermined threshold. Each entry is assigned a weight, which determines its importance in the overall calculation. The neuron adds the weighted inputs and, if the sum exceeds a certain threshold value, fires and produces an output of 1; otherwise, it remains idle and produces an output of 0. This binary output can be used as input to other neurons or as the final output of the neural network.

McCulloch and Pitts showed that connecting these artificial neurons in specific ways could perform complex calculations. They showed that networks of M-P neurons could compute logical functions, such as logical conjunction (DNA) and logical disjunction (OR). They also showed that these networks could be combined to perform more sophisticated calculations, such as universal computation, in which any computable function could be represented using a suitable network configuration.

Although the M-P neuron model was simplistic compared to biological neurons, it provided a crucial theoretical framework for understanding neural computation and inspired later developments in neural network research. His work paved the way for the development of more advanced artificial neural network models, including the perceptron and modern deep learning architectures used today.

The origin of everything: 1949

“Giant Brains or Thinking Machines” is a book written by Edmund Callis Berkley, first published in 1949. Explore the field of artificial intelligence (AI) and the concept of building machines that possess human-like thinking capabilities. Despite being written several decades ago, the book is still relevant, as it delves into fundamental concepts and challenges of AI.

Berkley begins with an introduction to the history of computers, early developments, and the possibility of machines simulating human intelligence. It then delves into the theoretical foundations of AI, covering topics such as logic, decision making and learning. It explores the idea of using electronic circuits to simulate the functions of the human brain, highlighting the advances made in electronic computing machines of the time.

The author also delves into the possible applications of AI in various fields, such as medicine, engineering, and even the arts. Berkley explores the ethical implications of creating machines capable of thinking and their potential impact on society. It raises questions about the future of work and the relationship between humans and intelligent machines.

Although “Giant Brains or Thinking Machines” was written at a time when the field of AI was in its early stages, it provides a fundamental understanding of the subject. The book presents the author’s view on intelligent machines and raises important questions about the possibilities and implications of creating machines that can think.

It should be noted that the book was written before the advent of modern AI techniques such as neural networks and deep learning, so some of the technical details and predictions made in the book may be out of date. However, “Giant Brains or Thinking Machines” remains a suggestive exploration of the concept of artificial intelligence and its possible impact on society.

Alan Turing: 1950

“Computer Machinery and Intelligence” is an influential essay written by Alan Turing, a pioneering figure in the field of computer science and artificial intelligence. Originally published in 1950, the essay addresses the question of whether machines can display true intelligence and proposes what is now known as the “Turing test” as a measure of machine intelligence.

Turing begins by questioning the notion of defining intelligence in absolute terms. It argues that intelligence should be evaluated based on the ability to display intelligent behavior and not on internal mental states or mechanisms. This lays the groundwork for his exploration of the potential of machines to think.

Turing proposes a thought experiment called the “imitation game,” which would later become known as the Turing test. In this game, an interrogator interacts with a human and a machine through a computer terminal, with the aim of determining which is the human and which is the machine. Turing suggests that if a machine can successfully trick the interrogator into thinking it is human, then it can be considered to possess intelligence.

The essay discusses possible objections and counterarguments to the idea of machine intelligence, including the “consciousness argument” and the limitations of computing machinery. Turing addresses these objections and postulates that the ability to think is not unique to humans, but can be achieved by machines through proper programming.

Turing also reflects on the possible social repercussions of artificial intelligence, addressing issues related to unemployment and the relationship between humans and machines. It raises philosophical questions about the nature of human intelligence and the potential of machine consciousness.

“Computer Machinery and Intelligence” is a fundamental work that laid the foundations of the field of artificial intelligence. Turing’s exploration of imitation play and his arguments about machine intelligence continue to shape the current development and understanding of AI. The essay remains highly influential and is considered a key contribution to the philosophical and practical discourse around artificial intelligence and the nature of intelligence itself.

Arthur Samuel: the father of machine learning

Samuel pioneered the development of a program that could play checkers, a project begun in 1952. This revolutionary achievement not only demonstrated the potential of AI, but also marked a major turning point in the field of gaming algorithms.

Samuel’s pioneering work on checkers culminated in 1955 with the creation of his program, known as the Samuel Checkers-Playing Program. Unlike previous approaches, which relied on exhaustive search algorithms, Samuel’s program incorporated machine learning techniques. Using a heuristic approach, the program learned from previous games, evaluated board positions, and adjusted its strategy based on experience. This was a significant change from traditional programming, as the program was now able to improve its game over time.

Dartmouth Workshop and Early AI Concepts: 1956

The Dartmouth Workshop, held in the summer of 1956, is a landmark event that marked the birth of artificial intelligence (AI) as an academic discipline. This groundbreaking gathering brought together influential figures in the field who laid the groundwork for the development of AI.

The Birth of AI: History of Artificial Intelligence

In the summer of 1956, John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon organized a seminal event known as the Dartmouth Workshop. The aim of this workshop was to explore the possibility of creating machines that could display human-like intelligence. Participants, who included mathematicians, computer scientists and cognitive psychologists, met for eight weeks of intense brainstorming and collaboration.

The Dartmouth Workshop proved to be a catalyst for AI research and laid the groundwork for the future development of this field. Participants were optimistic about the potential of AI and foresaw the creation of machines capable of solving complex problems, understanding natural language and even learning from experience. Although some of these goals were ambitious for the time, the workshop lit a spark that led to decades of pioneering research and innovation in AI.

Early AI Concepts: Symbolic AI and Logical Reasoning

During the Dartmouth Workshop, early AI researchers focused on symbolic AI, which aimed to develop intelligent systems by using symbolic representation and logical reasoning. Allen Newell and Herbert Simon’s Logic Theorist program, developed in 1955, was one of the first significant achievements in this direction. The program showed that machines could prove mathematical theorems by applying logical rules.

Another notable concept that emerged from the workshop was the idea of the “general problem solver.” Newell, Simon, and J.C. Shaw proposed the development of a universal problem-solving machine that could address a wide range of problems using a set of general-purpose heuristics. While implementing a comprehensive and comprehensive problem solver was challenging, the concept laid the foundation for further advances in AI problem-solving techniques.

The Dartmouth Workshop also explored the notion of machine learning. Although early AI researchers were optimistic about its potential, advances in this field were relatively limited at the time. Limitations in computing power and available data made it difficult to develop practical machine learning algorithms.

Despite these limitations, the Dartmouth Workshop provided a platform for fruitful discussions, exchange of ideas, and formulation of fundamental concepts that shaped the future trajectory of AI research. The event not only laid the groundwork for subsequent advances, but also fostered a vibrant community of researchers who continued to explore and push the boundaries of AI.

Logical theorist

In 1956, Newell and Simon, along with their colleague J.C. Shaw, introduced Logical Theorist as a program designed to mimic human problem-solving and reasoning processes. His goal was to demonstrate that a computer program could emulate human intelligence in the context of proving mathematical theorems.

Logic Theorist was developed for the IBM 704 computer and used symbolic logic and a set of heuristic rules to search for evidence. It worked by generating chains of logical implications and comparing them to a given set of axioms and theorems. The program employed a combination of forward and backward chains to explore and manipulate logical expressions, attempting to find a proof for a given theorem.

One of the most notable achievements of Logic Theorist was to successfully prove 38 of the 52 theorems of symbolic logic from Principia Mathematica, a monumental work by mathematicians Alfred North Whitehead and Bertrand Russell. The Logic Theorist’s ability to prove theorems demonstrated that automated reasoning was possible and opened up new possibilities for computer-assisted mathematical exploration.

The development of the Logical Theorist was an important milestone in the field of artificial intelligence, as it demonstrated the potential of computers to perform tasks traditionally associated with human intelligence. Newell and Simon’s work paved the way for research and development in the field of automated theorem proving and symbolic reasoning, and their ideas continue to influence this field to this day.

The Birth of Machine Learning: Perceptrons

In 1957, the field of machine learning experienced a breakthrough with the introduction of the perceptron by Frank Rosenblatt. The perceptron was an early form of artificial neural network inspired by the workings of the human brain. He was able to learn from data through a process known as supervised learning.

The perceptron demonstrated remarkable capabilities in pattern recognition tasks, highlighting its potential for machine learning. It received a great deal of attention and raised hopes of developing more advanced intelligent systems.

However, the initial enthusiasm around perceptrons was short-lived due to their limitations. Perceptrons could only classify linearly separable patterns, which meant they struggled with more complex problems that required nonlinear decision limits.


Joseph Weizenbaum, a computer scientist and professor at MIT, developed ELIZA, a pioneering computer program that simulated a psychotherapist. ELIZA, created in the mid-60s, is considered one of the first examples of natural language processing and artificial intelligence.

ELIZA aimed to simulate a conversation between a user and a psychotherapist using a technique called “pattern matching.” The program analyzed user input and generated responses based on predefined patterns and rules. ELIZA employed a simple but effective method of transforming user statements into questions and reflections, making it appear that the program understood and engaged in meaningful conversation.

The key tenet of ELIZA was the idea of “Rogerian psychotherapy,” developed by psychologist Carl Rogers. Rogerian therapy emphasizes active listening, empathy, and reflection, and the therapist encourages the client to explore their thoughts and emotions. ELIZA incorporated these techniques by mirroring users’ statements and asking open-ended questions, without providing genuine understanding or emotional insight.

ELIZA gained attention and popularity for its ability to engage users and create the illusion of meaningful conversation. Weizenbaum’s intention was to demonstrate the superficiality of human-computer interaction and to challenge the idea that computers could truly understand or empathize with humans.

Despite its limitations and lack of true intelligence, ELIZA had a significant impact in the field of artificial intelligence and natural language processing. It sparked new research and inspired the development of more sophisticated chatbot systems. ELIZA’s influence can be seen in contemporary conversational agents, including virtual assistants and chatbots, which continue to evolve with advances in AI technologies.


In the 1970s, AI research hit a roadblock. The field was plagued by problems such as the “knowledge representation problem” and the “framework problem.” These problems made it difficult for AI systems to represent knowledge and reason about the world in a similar way to humans.

In the field of artificial intelligence (AI), the problem of knowledge representation refers to the challenge of effectively representing and organizing knowledge within a computational system. It’s about finding appropriate ways to store, structure and manipulate knowledge so that an AI system can reason, learn and make intelligent decisions.

The problem of knowledge representation arises from the fact that human knowledge is vast, diverse, and often complex. Translating this knowledge into a format that machines can understand and use is a fundamental challenge of AI.

The framework problem is a well-known issue in artificial intelligence (AI) and represents a challenge related to representation and reasoning about the effects of actions and changes in dynamic environments. It was first identified by AI researchers John McCarthy and Patrick J. Hayes in the late 1960s.

The problem of the framework arises from the difficulty of determining which aspects of a situation are relevant to represent them explicitly and update them when considering the effects of an action, ignoring those that remain unchanged. In other words, it’s about the problem of representing and reasoning about what stays the same (or doesn’t change) when something changes.

To illustrate this problem, let’s look at a simple example: Suppose an AI system is tasked with changing the state of a room by opening a window. The system has to recognize what information is relevant to describe the initial state of the room, what changes need to be made and what remains the same.

The framework problem highlights the challenge of avoiding overcalculations and unnecessary updates of the entire knowledge base every time an action is taken. The problem lies in distinguishing the relevant changes from the large amount of background knowledge that remains unchanged.

Researchers have proposed several solutions to address the framework problem. Some approaches involve explicit representations of the effects of actions, including the introduction of action-specific rules or logical axioms. Others use more implicit methods, such as default reasoning or non-monotonic logic, to solve the frame problem.

Despite these attempts, the framework problem remains an unfinished challenge in AI. It is closely related to the broader question of representation and reasoning about change, and remains an active area of research, especially in the fields of knowledge representation, planning, and reasoning about action and time in AI systems.

Stumbles and to make matters worse, little funding.


The 80s were a decade of great advances in artificial intelligence (AI). After a period of initial enthusiasm in the sixties and seventies, AI research had stalled somewhat by the early eighties. However, a series of new breakthroughs in the 1980s helped revive AI research and usher in a new era of progress.

One of the most important advances of the 80s was the rise of machine learning. Machine learning is an artificial intelligence technique that allows computers to acquire knowledge from data, eliminating the need for explicit programming, which was a breakthrough as it opened up a whole new world of possibilities for AI.

Another important advance of the 80s was the development of expert systems. Expert systems are computer programs that can mimic the decision-making process of a human expert. This was a breakthrough, as it allowed AI to be used for the first time in real-world applications.

The 1980s also saw the development of other important AI technologies, such as natural language processing and computer vision. These technologies laid the foundation for the even faster progress that would occur in AI in the 1990s and beyond.

Below is a timeline of some of the major advances in AI in the 1980s:

  • 1980: Development of WABOT-2, a humanoid robot capable of interacting with people, reading sheet music and playing music on an electronic organ.
  • 1982: Launch of the Fifth Generation Computer Systems project, a Japanese government initiative to develop a new generation of computers capable of human-level reasoning.
  • 1983: Development of the Dendral expert system, capable of diagnosing diseases based on medical symptoms.
  • 1984: Premiere of the film “2001: A Space Odyssey”, in which a HAL 9000 computer appears that becomes aware of itself and becomes homicidal.
  • 1985: Development of the expert system MYCIN, capable of diagnosing infectious diseases.
  • 1986: Development of the R1 expert system, capable of configuring computer systems.
  • 1987: Development of the PROLOG programming language, used for logic programming.
  • 1988: Development of the Soar cognitive architecture, which is a model of human cognition.
  • 1989: Development of the Neural Network programming language, used for neural network programming.

The 80s were a decade of great advances in AI. The development of new technologies, such as machine learning, expert systems, and natural language processing, laid the groundwork for the even faster progress that would occur in AI in the 1990s and beyond.


The rise of statistical learning and data-driven approaches: In the 1990s, machine learning shifted toward statistical approaches, emphasizing the analysis of data and patterns. The researchers explored techniques such as decision trees, support vector machines (SVMs), and Bayesian networks. In this period also emerged the field of data mining, whose objective was to extract useful information and knowledge from large data sets.

The 90s were a decade of great advances in artificial intelligence (AI). After a period of rapid growth in the 1980s, AI research continued to advance in the 90s, achieving a number of major breakthroughs.

One of the most important advances of the 1990s was the rise of deep learning. Deep learning, a branch of machine learning, harnesses the power of artificial neural networks to extract insights and insights from data. This was a breakthrough, as it allowed AI to achieve superhuman performance in a number of tasks, such as image recognition and natural language processing.

Another important advance of the 90s was the development of autonomous robots. Autonomous robots are characterized by their ability to operate independently, without the need for human intervention. They are able to carry out tasks and make decisions based on their own programming and sensory information, eliminating dependence on constant human control, which was a breakthrough as it opened up a whole new world of possibilities for AI, for example in the fields of manufacturing, healthcare and transport.

The 1990s also saw the development of other important AI technologies, such as genetic algorithms and evolutionary computation. These technologies laid the foundation for the even faster progress that would occur in AI in the 2000s and beyond.

Below is a timeline of some of the major advances in AI in the 1990s:

  • 1990: Development of the Deep Blue chess program, which defeated world champion Garry Kasparov in 1997.
  • 1991: Development of ARPANET, precursor to the Internet, which enabled AI researchers to share data and collaborate on projects.
  • 1992: Development of the Dactyl hand, a robotic hand capable of grasping and manipulating objects.
  • 1993: Development of the Shaky robot, a mobile robot capable of moving around its environment and avoiding obstacles.
  • 1994: Development of the Watson natural language processing system, capable of understanding and responding to human language.
  • 1995: Development of the robot dog Aibo, one of the first commercially successful robots.
  • 1996: Development of the ImageNet dataset, a large collection of images used to train machine learning models.
  • 1997: Development of the LeNet convolutional neural network, one of the first deep learning models to achieve success in image recognition.
  • 1998: Development of the GeneXpert diagnostic system, which uses genetic algorithms to diagnose infectious diseases.


The 2000s were a decade of great advances in artificial intelligence (AI). After a period of rapid growth in the 90s, AI research continued to advance in the 2000s, in which several important advances were made.

One of the most important advances of the 2000s was the rise of deep learning. Deep learning, a form of machine learning, employs artificial neural networks to gain insights and insights from data. Leveraging these intricate networks, deep learning algorithms can autonomously analyze and understand complex patterns, facilitating advanced data processing and learning capabilities. This was a breakthrough, as it allowed AI to achieve superhuman performance in a number of tasks, such as image recognition and natural language processing.

Another major breakthrough of the 2000s was the development of autonomous robots. The development of autonomous robots was a breakthrough in artificial intelligence. These robots can operate without human intervention, opening up a whole new world of possibilities for AI in various fields. For example, autonomous robots can be used in manufacturing to perform dangerous or repetitive tasks, in healthcare to assist patients and caregivers, and in transportation to deliver goods and services.

The 2000s also saw the development of other important AI technologies, such as genetic algorithms and evolutionary computation. These technologies laid the foundation for the even faster progress that would occur in AI in the 2010s and beyond.

Here’s a timeline of some of the major advances in AI in the 2000s:

  • 2000: The development of the Nomad robot, which scans remote regions of Antarctica for meteorite samples.
  • 2002: The development of iRobot Roomba, which autonomously vacuums the ground while navigating and avoids obstacles.
  • 2004: Development of the OWL web ontology language, which is used to represent knowledge in a machine-readable format.
  • 2005: Development of the DARPA Grand Challenge, a competition to develop autonomous vehicles that can traverse the desert without human intervention.
  • 2006: The development of the Google Translate service, which can translate texts from one language to another.
  • 2009: Development of the DeepMind artificial intelligence system, which can learn to play Atari games at superhuman levels.



  • IBM’s Watson supercomputer wins the Jeopardy contest, demonstrating AI’s ability to understand natural language and answer complex questions.


  • Apple introduces Siri, an intelligent personal assistant for iOS devices, popularizing voice-controlled AI apps.


  • Google’s DeepMind is developing a deep neural network called AlexNet, which makes a breakthrough in image recognition accuracy and kicks off the resurgence of deep learning.


  • DeepMind’s AlphaGo defeats a human Go world champion for the first time, demonstrating the potential of AI in complex strategic games.


  • OpenAI, a non-profit research organization, is founded with a mission to ensure that artificial general intelligence (AGI) benefits all of humanity.


Generative adversarial networks (GANs) attract attention for their ability to generate realistic images and videos.
Tesla announces its Autopilot feature, which uses artificial intelligence and machine learning algorithms to enable advanced driver assistance features in its vehicles.


The European Union introduces the General Data Protection Regulation (GDPR) to address data privacy concerns and establish guidelines for AI applications.
OpenAI publishes GPT-2, a large-scale linguistic model capable of generating coherent and contextually relevant text.


In 2020, the COVID-19 pandemic accelerated the development and adoption of AI technologies. AI was used to develop new diagnostic tools, track the spread of the virus, and develop vaccines. For example, AI was used to develop the chest X-ray analysis tool that was used to identify COVID-19 patients. It was also used to develop the contact tracing apps that were used to track the spread of the virus. And AI was used to develop the vaccine development tools.


In 2021, the OpenAI GPT-3 linguistic model was published. GPT-3 is a great linguistic model capable of generating human-quality text. It can be used to create realistic-looking articles, social media posts, and even poems. GPT-3 has been used to create various products and services, such as a chatbot capable of answering customer questions, a tool capable of generating creative content, and a system capable of translating languages.


ChatGPT 3.5 is released this year; He’s trained on a massive data set of text and code, allowing him to generate human-quality text, translate languages, write different types of creative content, and answer your questions in an informative way.


In 2023, Google launched LaMDA AI. The MDA AI is a linguistic model capable of having conversations indistinguishable from those held with a human. MDA AI represents a significant advance over previous linguistic models, as it is able to understand and respond to complex questions and requests.

Also this year debuted Chat GPT4, a much more advanced and powerful model than GPT-3.5.

Read also: History of digital marketing, evolution, timeline, chronology; Fifth generation computers, 1980 onwards, features; Data Center Evolution, History; Algorithm Meaning; Xiaomi humanoid robot; What is Ezoic?

Valuable external resources: Tableau; harvard.edu;

This post is also available in: English Deutsch (German) Español (Spanish) Nederlands (Dutch)