ARTIFICIAL INTELLIGENCE
AI
AI classifies machines that mimic human intelligence and cognitive functions, such as problem-solving and learning. It utilizes predictions and automation to optimize and solve complex tasks.
SELECT/
UNSELECT
TO FILTER .REFRESH THE PAGE FOR REGULAR VIEW.
Beyond the Timeline
Three types of AI are commonly distinguished. Narrow or Weak AI (ANI) is designed for specific tasks, such as automating processes, improving efficiency and analyzing complex data, often surpassing human speed and accuracy within its narrow scope; all AI systems in use today are generally classified as ANI. General or Strong AI (AGI) is a theoretical form of AI with human-level cognitive abilities across domains. Super AI (ASI) is a hypothetical future AI that would surpass human intelligence.
AI as a concept dates back to ancient times, when inventors created "automatons," devices that moved independently of human intervention. One of the earliest recorded was a mechanical pigeon attributed to Archytas of Tarentum, a friend of Plato, said to have been powered by steam or compressed air to mimic a dove's flight. Leonardo da Vinci's notebooks include designs for a mechanical knight around 1495, centuries before the word "robot" existed. Around 1560, the Italo-Spanish clockmaker, engineer, and mathematician Juanelo Turriano of Toledo reportedly created a 15-inch wooden-and-iron automaton of a monk.
In 1950, mathematician Alan Turing published Computing Machinery and Intelligence, asking: can machines think? He proposed an "imitation game," now known as the Turing test, in which a machine is considered intelligent if a human judge cannot reliably tell it apart from a person.
The term "artificial intelligence" was coined in the 1955 proposal for the Dartmouth Workshop, held in the summer of 1956, which established AI as a field of research. In 1958, psychologist Frank Rosenblatt introduced the Perceptron, one of the first artificial neural networks able to learn from experience. In 1969, Marvin Minsky and Seymour Papert published an analysis of its limitations, and reduced funding in the 1970s led to the first AI Winter.
Artificial neural networks, the foundation of deep learning, trace back to 1943, when Warren McCulloch and Walter Pitts showed that simplified models of neurons could perform logical functions. These artificial neurons were loosely inspired by those of the human brain. Key milestones that followed include the Perceptron, backpropagation, and AlexNet's 2012 win in the ImageNet challenge.
During the 1960s and 1970s, AI research focused mainly on rule-based systems and symbolic reasoning, producing programs that proved theorems, held simple conversations, such as ELIZA, and played games such as checkers. Many were written in LISP, the language John McCarthy created in 1958.
In 1972, Stanford researchers began developing MYCIN, an expert system that recommended antibiotics for bacterial infections. In the mid-1970s, limited computing power and reduced funding led to the first AI Winter, a period of declining interest and investment.
Machine learning, in which computers learn from data instead of following explicitly programmed rules, dates to 1959, when IBM researcher Arthur Samuel coined the term while building a checkers program that improved as it played. In 1986, David Rumelhart, Geoffrey Hinton and Ronald Williams popularized backpropagation, a method for training multi-layer neural networks that became a foundation of modern deep learning.
In the 1990s, support vector machines (SVMs), based on work by Vladimir Vapnik and Alexey Chervonenkis, became widely used for classification problems. Decision trees also became popular as interpretable models for classification and regression, and laid the groundwork for ensemble methods, which improved predictive performance. AI was applied to fraud detection, document classification, and face recognition, and Q-learning, introduced in 1989, was refined to handle more complex decision-making problems. In 1997, IBM's Deep Blue defeated world chess champion Garry Kasparov in a six-game match.
In the 2000s, faster hardware and larger datasets revived research into neural networks with many layers, known as deep learning. In 2011, IBM's Watson defeated former champions Ken Jennings and Brad Rutter on Jeopardy!. In 2016, DeepMind's AlphaGo defeated Lee Sedol, one of the world's top Go players, four games to one. Other milestones of the period include robotic vacuum cleaners, self-driving cars on public roads, and virtual assistants such as Siri and Alexa. In November 2022, OpenAI released ChatGPT, which reached 100 million monthly active users within two months.
The use of AI algorithms in critical areas such as hiring and criminal justice has raised questions about bias and de-skilling. AI-generated synthetic media, known as "deepfakes," has been used to spread false information, defame individuals, and attempt to influence elections.
Another concern is that AI systems could eventually exceed human understanding or control, an issue that has become part of US policy debates about how to govern the technology.
AI is being integrated into industries such as healthcare, finance, and transportation. In medicine, AI systems analyze medical images and patient data to help doctors with diagnoses, including systems designed to detect heart disease, cancer, and eye disease. Self-driving cars use sensors, cameras, and AI software to navigate roads without human intervention.
AI infrastructure refers to the software and hardware needed to develop and deploy AI-powered applications. It includes MLOps platforms, compute resources, ML frameworks, and data storage and processing solutions.
In 2023, Bloomberg Intelligence projected that generative AI could grow from a $40 billion market in 2022 to $1.3 trillion by 2032, and that it could add about $280 billion in new software revenue, driven by specialized assistants, new infrastructure products, and copilots that accelerate coding.
Microsoft first invested in OpenAI in 2019 and expanded the partnership in January 2023, with Microsoft Azure providing OpenAI's cloud computing. On October 28, 2025, OpenAI completed a restructuring into a for-profit public benefit corporation, OpenAI Group PBC, controlled by the nonprofit OpenAI Foundation. Under the new agreement, Microsoft holds about 27% of OpenAI, a stake valued at approximately $135 billion, and keeps rights to use OpenAI's technology through 2032. OpenAI committed to purchasing $250 billion of Azure services, while Microsoft gave up its right of first refusal to be OpenAI's cloud provider.
In January 2025, OpenAI announced the Stargate Project, a new company that intends to invest $500 billion over four years in AI infrastructure in the United States, beginning with $100 billion. Its equity funders are SoftBank, OpenAI, Oracle and Abu Dhabi's MGX. The flagship campus in Abilene, Texas, has had two buildings operating since September 2025.
Add a Title
Beyond the Timeline
Three types of AI are commonly distinguished. Narrow or Weak AI (ANI) is designed for specific tasks, such as automating processes, improving efficiency and analyzing complex data, often surpassing human speed and accuracy within its narrow scope; all AI systems in use today are generally classified as ANI. General or Strong AI (AGI) is a theoretical form of AI with human-level cognitive abilities across domains. Super AI (ASI) is a hypothetical future AI that would surpass human intelligence.
AI as a concept dates back to ancient times, when inventors created "automatons," devices that moved independently of human intervention. One of the earliest recorded was a mechanical pigeon attributed to Archytas of Tarentum, a friend of Plato, said to have been powered by steam or compressed air to mimic a dove's flight. Leonardo da Vinci's notebooks include designs for a mechanical knight around 1495, centuries before the word "robot" existed. Around 1560, the Italo-Spanish clockmaker, engineer, and mathematician Juanelo Turriano of Toledo reportedly created a 15-inch wooden-and-iron automaton of a monk.
In 1950, mathematician Alan Turing published Computing Machinery and Intelligence, asking: can machines think? He proposed an "imitation game," now known as the Turing test, in which a machine is considered intelligent if a human judge cannot reliably tell it apart from a person.
The term "artificial intelligence" was coined in the 1955 proposal for the Dartmouth Workshop, held in the summer of 1956, which established AI as a field of research. In 1958, psychologist Frank Rosenblatt introduced the Perceptron, one of the first artificial neural networks able to learn from experience. In 1969, Marvin Minsky and Seymour Papert published an analysis of its limitations, and reduced funding in the 1970s led to the first AI Winter.
Artificial neural networks, the foundation of deep learning, trace back to 1943, when Warren McCulloch and Walter Pitts showed that simplified models of neurons could perform logical functions. These artificial neurons were loosely inspired by those of the human brain. Key milestones that followed include the Perceptron, backpropagation, and AlexNet's 2012 win in the ImageNet challenge.
During the 1960s and 1970s, AI research focused mainly on rule-based systems and symbolic reasoning, producing programs that proved theorems, held simple conversations, such as ELIZA, and played games such as checkers. Many were written in LISP, the language John McCarthy created in 1958.
In 1972, Stanford researchers began developing MYCIN, an expert system that recommended antibiotics for bacterial infections. In the mid-1970s, limited computing power and reduced funding led to the first AI Winter, a period of declining interest and investment.
Machine learning, in which computers learn from data instead of following explicitly programmed rules, dates to 1959, when IBM researcher Arthur Samuel coined the term while building a checkers program that improved as it played. In 1986, David Rumelhart, Geoffrey Hinton and Ronald Williams popularized backpropagation, a method for training multi-layer neural networks that became a foundation of modern deep learning.
In the 1990s, support vector machines (SVMs), based on work by Vladimir Vapnik and Alexey Chervonenkis, became widely used for classification problems. Decision trees also became popular as interpretable models for classification and regression, and laid the groundwork for ensemble methods, which improved predictive performance. AI was applied to fraud detection, document classification, and face recognition, and Q-learning, introduced in 1989, was refined to handle more complex decision-making problems. In 1997, IBM's Deep Blue defeated world chess champion Garry Kasparov in a six-game match.
In the 2000s, faster hardware and larger datasets revived research into neural networks with many layers, known as deep learning. In 2011, IBM's Watson defeated former champions Ken Jennings and Brad Rutter on Jeopardy!. In 2016, DeepMind's AlphaGo defeated Lee Sedol, one of the world's top Go players, four games to one. Other milestones of the period include robotic vacuum cleaners, self-driving cars on public roads, and virtual assistants such as Siri and Alexa. In November 2022, OpenAI released ChatGPT, which reached 100 million monthly active users within two months.
The use of AI algorithms in critical areas such as hiring and criminal justice has raised questions about bias and de-skilling. AI-generated synthetic media, known as "deepfakes," has been used to spread false information, defame individuals, and attempt to influence elections.
Another concern is that AI systems could eventually exceed human understanding or control, an issue that has become part of US policy debates about how to govern the technology.
AI is being integrated into industries such as healthcare, finance, and transportation. In medicine, AI systems analyze medical images and patient data to help doctors with diagnoses, including systems designed to detect heart disease, cancer, and eye disease. Self-driving cars use sensors, cameras, and AI software to navigate roads without human intervention.
AI infrastructure refers to the software and hardware needed to develop and deploy AI-powered applications. It includes MLOps platforms, compute resources, ML frameworks, and data storage and processing solutions.
In 2023, Bloomberg Intelligence projected that generative AI could grow from a $40 billion market in 2022 to $1.3 trillion by 2032, and that it could add about $280 billion in new software revenue, driven by specialized assistants, new infrastructure products, and copilots that accelerate coding.
Microsoft first invested in OpenAI in 2019 and expanded the partnership in January 2023, with Microsoft Azure providing OpenAI's cloud computing. On October 28, 2025, OpenAI completed a restructuring into a for-profit public benefit corporation, OpenAI Group PBC, controlled by the nonprofit OpenAI Foundation. Under the new agreement, Microsoft holds about 27% of OpenAI, a stake valued at approximately $135 billion, and keeps rights to use OpenAI's technology through 2032. OpenAI committed to purchasing $250 billion of Azure services, while Microsoft gave up its right of first refusal to be OpenAI's cloud provider.
In January 2025, OpenAI announced the Stargate Project, a new company that intends to invest $500 billion over four years in AI infrastructure in the United States, beginning with $100 billion. Its equity funders are SoftBank, OpenAI, Oracle and Abu Dhabi's MGX. The flagship campus in Abilene, Texas, has had two buildings operating since September 2025.