History of AI
The concept of artificial intelligence began to gain popularity in the 1950s when computer scientist Alan Turing published a paper called “Computing Machinery and Intelligence,” which asked the question of whether machines could think and how to test their intelligence. This paper laid the foundation for AI research and development and was the first to propose the Turing test, a method for assessing machine intelligence. The term “artificial intelligence” was coined by computer scientist John McCarthy in 1956 at a Dartmouth College academic conference.
Following the McCarthy conference, interest in AI research grew throughout the 1970s, both at academic institutions and funded by the U.S. government. Innovations in computing established several AI foundations during this time, including machine learning, neural networks, and natural language processing. Despite the advancements, AI technology eventually became harder to scale than expected, leading to a decline in interest and funding, and the first AI winter lasted until the 1980s.
In the mid-1980s, interest in AI was rekindled by improved computer power, the popularity of deep learning, and the introduction of AI-powered “expert systems.” However, the complexity of the new systems and the inability of existing technology to keep up led to a second AI winter that lasted until the mid-1990s.
By the mid-2000s, innovations in processing power, big data, and advanced deep learning techniques removed previous barriers to AI, enabling further breakthroughs in AI. Modern AI technologies such as virtual assistants, self-driving cars, and generative AI began to enter the mainstream in the 2010s and have shaped AI as we know it today.
Artificial Intelligence Timeline
(1943) Warren McCullough and Walter Pitts published their paper, “The Logical Calculus of Ideas Inherent in Neural Activity,” proposing the first mathematical model for building neural networks.
(1949) In his book The Organization of Behavior: A Neuropsychological Theory, Donald Hebb proposes that neural pathways are created by experience, and the connections between neurons become stronger the more frequently they are used. Hebbian learning continues to be an important model in AI.
(1950) Alan Turing published his paper “Computing Machinery and Intelligence” in which he proposed what is now known as the Turing test as a way to determine whether a machine is intelligent.
(1950) Harvard undergraduates Marvin Minsky and Dean Edmonds develop the first neural network computer, SNARC.
(1956) The term “artificial intelligence” was coined at Dartmouth College's Summer Research Project on Artificial Intelligence. Led by John McCarthy, the conference is widely considered the birthplace of AI.
(1958) John McCarthy developed the AI programming language Lisp and published a paper called “Programs with Common Sense”, which proposes a hypothetical Advice Taker, a complete AI system capable of learning from experience as effectively as humans.
(1959) Arthur Samuel coined the term “machine learning” while at IBM.
(1964) As a doctoral student at MIT, Daniel Bobrow developed STUDENT, an early natural language processing program designed to solve algebra word problems.
(1966) MIT professor Joseph Weizenbaum created Eliza, one of the first chatbots to successfully mimic users' conversational patterns, creating the illusion of understanding them better than they actually did. This gave rise to the Eliza effect, a common phenomenon in which people mistakenly assume that AI systems have human-like thought processes and emotions.
(1969) The first successful expert systems, DENDRAL and MYCIN, were created at Stanford University’s AI Lab.
(1972) The logic programming language PROLOG is developed.
(1973) The Lighthill Report was published by the UK government detailing disappointments in AI research, leading to major cuts in funding for AI projects.
(1974-1980) Frustrated with the progress of AI development, DARPA made drastic cuts to academic grant funding. Combined with the earlier ALPAC report and the Lighthill report the previous year, AI funding dried up and research stagnated. This period is known as the “First AI Winter.”
(1980) Digital Equipment Corporations developed the first successful commercial expert system, R1 (also known as XCON). Designed for configuring orders for new computer systems, R1 sparked a boom in investment in expert systems that lasted for the better part of a decade, effectively ending the first AI winter.
(1985) Companies spend more than $1 billion a year on expert systems, and an entire industry known as the Lisp machine market has sprung up to support them: Companies such as Symbolics and Lisp Machines Inc. build specialized computers that run on the AI programming language Lisp.
(1987-1993) As computing technology improved, cheaper alternatives emerged, leading to the collapse of the Lisp machine market in 1987, ushering in the “Second AI Winter.” During this period, expert systems proved too expensive to maintain and update, and eventually fell out of favor.
(1997) IBM's Deep Blue defeats World Chess Champion Garry Kasparov.
(2006) Fei-Fei Li started developing the ImageNet visual database, which was introduced in 2009, which sparked the AI boom and laid the foundation for the development of image recognition.
(2008) Google has made a breakthrough in voice recognition and has introduced the feature to its iPhone app.
(2011) IBM's Watson handily beats the competition on Jeopardy!
(2011) Apple has released its AI-powered virtual assistant, Siri, through its iOS operating system.
(2012) Andrew Ng, founder of the Google Brain Deep Learning project, fed a neural network using a deep learning algorithm 10 million YouTube videos as a training set. The neural network learned to recognize cats without knowing what cats were, ushering in a groundbreaking era of neural network and deep learning funding.
(2014) The virtual home smart device, Amazon Alexa, has been released.
(2016) Google DeepMind's AlphaGo defeats world champion Go player Lee Sedol, whose complexity in this ancient Chinese game was seen as a major hurdle for AI to overcome.
(2018) Google releases BERT, a natural language processing engine, to lower the barrier to translation and understanding through ML applications.
(2020) Baidu released its LinearFold AI algorithm to scientific and medical teams working on vaccine development in the early stages of the SARS-CoV-2 pandemic. The algorithm could predict the virus' RNA sequence in just 27 seconds, 120 times faster than other methods.
(2020) OpenAI has released GPT-3, a natural language processing model that can generate text modeled on the way humans speak and write.
(2021) Building on GPT-3, OpenAI is developing DALL-E, which can create images from text prompts.
(2022) The National Institute of Standards and Technology has published the first edition of its AI Risk Management Framework, a set of voluntary U.S. guidelines to “better manage the risks to individuals, organizations, and society associated with artificial intelligence.”
(2022) OpenAI released ChatGPT, a chatbot with a large language model that has gained over 100 million users in just a few months.
(2022) The White House has introduced an AI Bill of Rights outlining principles for the responsible development and use of AI.
(2023) Microsoft has released an AI-powered version of its search engine Bing, built on the same technology as ChatGPT.
(2023) Google launches Bard, a competing conversational AI that later becomes Gemini.
(2023) OpenAI has unveiled GPT-4, the most sophisticated language model to date.
(2023) The Biden-Harris Administration issued an Executive Order on Safe, Secure, and Trustworthy AI, calling for increased efforts to test for safety, label AI-generated content, and develop international standards for the development and use of AI. The order also emphasizes the importance of ensuring that artificial intelligence is not used to circumvent privacy protections, exacerbate discrimination, or violate civil and consumer rights.
(2023) Chatbot Grok It was released by Elon Musk's AI company xAI.
(2024) The European Union has passed an artificial intelligence bill that aims to ensure that AI systems deployed within the EU are “safe, transparent, traceable, non-discriminatory and environmentally friendly.”
(2024) Claude 3 Opus, a large-scale language model developed by AI company Anthropic, outperformed GPT-4, becoming the first LLM to outperform GPT-4.
