Editing Artificial intelligence (section)

== History ==
{{Main|History of artificial intelligence}}
{{For timeline}}
[[File:2024 AI patents by country - artificial intelligence.svg |thumb |In 2024, AI patents in China and the US numbered more than three-fourths of AI patents worldwide.<ref name=RandDworld_20241103/> Though China had more AI patents, the US had 35% more patents per AI patent-applicant company than China.<ref name=RandDworld_20241103>{{cite web |last1=Buntz |first1=Brian |title=Quality vs. quantity: US and China chart different paths in global AI patent race in 2024 / Geographical breakdown of AI patents in 2024 |url=https://www.rdworldonline.com/quality-vs-quantity-us-and-china-chart-different-paths-in-global-ai-patent-race-in-2024/ |publisher=R&D World |archive-url=https://web.archive.org/web/20241209072113/https://www.rdworldonline.com/quality-vs-quantity-us-and-china-chart-different-paths-in-global-ai-patent-race-in-2024/ |archive-date=9 December 2024 |date=3 November 2024 |url-status=live}}</ref>]]
<!-- DON'T INCLUDE HISTORICAL PRECURSORS (THEY BELONG TO THE SEPARATE 'HISTORY OF AI' ARTICLE) -->
<!-- MAJOR INTELLECTUAL PRECURSORS: LOGIC, THEORY OF COMPUTATION, CYBERNETICS, INFORMATION THEORY, NEUROBIOLOGY, SPECULATION: Antiquity - 1955 -->
The study of mechanical or "formal" reasoning began with philosophers and mathematicians in antiquity. The study of logic led directly to [[Alan Turing]]'s [[theory of computation]], which suggested that a machine, by shuffling symbols as simple as "0" and "1", could simulate any conceivable form of mathematical reasoning.{{Sfn|Russell|Norvig|2021|p=9}}<ref name="Clarendon Press-2004"/> This, along with concurrent discoveries in [[cybernetics]], [[information theory]] and [[neurobiology]], led researchers to consider the possibility of building an "electronic brain".{{Efn|"Electronic brain" was the term used by the press around this time.{{Sfn|Russell|Norvig|2021|p=9}}<ref>{{Cite web |title=Google books ngram |url=https://books.google.com/ngrams/graph?content=electronic+brain&year_start=1930&year_end=2019&corpus=en-2019&smoothing=3 |access-date=5 October 2024 |archive-date=5 October 2024 |archive-url=https://web.archive.org/web/20241005170209/https://books.google.com/ngrams/graph?content=electronic+brain&year_start=1930&year_end=2019&corpus=en-2019&smoothing=3 |url-status=live }}</ref>}} They developed several areas of research that would become part of AI,<ref>AI's immediate precursors: {{Harvtxt|McCorduck|2004|pp=51–107}}, {{Harvtxt|Crevier|1993|pp=27–32}}, {{Harvtxt|Russell|Norvig|2021|pp=8–17}}, {{Harvtxt|Moravec|1988|p=3}}</ref> such as [[Warren McCullouch|McCullouch]] and [[Walter Pitts|Pitts]] design for "artificial neurons" in 1943,{{Sfnp|Russell|Norvig|2021|p=17}} and Turing's influential 1950 paper '[[Computing Machinery and Intelligence]]', which introduced the [[Turing test]] and showed that "machine intelligence" was plausible.<ref name="Turing"/><ref name="Clarendon Press-2004">{{Cite book |title=The Essential Turing: the ideas that gave birth to the computer age |date=2004 |publisher=Clarendon Press |isbn=0-1982-5079-7 |editor-last=Copeland |editor-first=J. |location=Oxford, England}}</ref>

<!-- 1956-1974 -->
The field of AI research was founded at [[Dartmouth workshop|a workshop]] at [[Dartmouth College]] in 1956.{{Efn|
Daniel Crevier wrote, "the conference is generally recognized as the official birthdate of the new science."{{Sfnp|Crevier|1993|pp=47–49}} [[Stuart J. Russell|Russell]] and [[Norvig]] called the conference "the inception of artificial intelligence."{{Sfnp|Russell|Norvig|2021|p=17}}}}<ref name="Dartmouth workshop">[[Dartmouth workshop]]: {{Harvtxt|Russell|Norvig|2021|p=18}}, {{Harvtxt|McCorduck|2004|pp=111–136}}, {{Harvtxt|NRC|1999|pp=200–201}}<br />The proposal: {{Harvtxt|McCarthy|Minsky|Rochester|Shannon|1955}}</ref> The attendees became the leaders of AI research in the 1960s.{{Efn|
[[Stuart J. Russell|Russell]] and [[Norvig]] wrote "for the next 20 years the field would be dominated by these people and their students."{{Sfnp|Russell|Norvig|2003|p=17}}
}} They and their students produced programs that the press described as "astonishing":{{Efn|
[[Stuart J. Russell|Russell]] and [[Norvig]] wrote, "it was astonishing whenever a computer did anything kind of smartish".{{Sfnp|Russell|Norvig|2003|p=18}}
}} computers were learning [[checkers]] strategies, solving word problems in algebra, proving [[Theorem|logical theorems]] and speaking English.{{Efn|
The programs described are [[Arthur Samuel (computer scientist)|Arthur Samuel]]'s checkers program for the [[IBM 701]], [[Daniel Bobrow]]'s [[STUDENT]], [[Allen Newell|Newell]] and [[Herbert A. Simon|Simon]]'s [[Logic Theorist]] and [[Terry Winograd]]'s [[SHRDLU]].
}}<ref name="Succ1">Successful programs of the 1960s: {{Harvtxt|McCorduck|2004|pp=243–252}}, {{Harvtxt|Crevier|1993|pp=52–107}}, {{Harvtxt|Moravec|1988|p=9}}, {{Harvtxt|Russell|Norvig|2021|pp=19–21}}</ref> Artificial intelligence laboratories were set up at a number of British and U.S. universities in the latter 1950s and early 1960s.<ref name="Clarendon Press-2004"/>

<!-- Optimism of the 60s and first AI "winter": 1974 -->
Researchers in the 1960s and the 1970s were convinced that their methods would eventually succeed in creating a machine with [[artificial general intelligence|general intelligence]] and considered this the goal of their field.{{Sfnp|Newquist|1994|pp=86–86}} In 1965 [[Herbert A. Simon|Herbert Simon]] predicted, "machines will be capable, within twenty years, of doing any work a man can do".<ref>{{Harvtxt|Simon|1965|p=96}} quoted in {{Harvtxt|Crevier|1993|p=109}}</ref> In 1967 [[Marvin Minsky]] agreed, writing that "within a generation&nbsp;... the problem of creating 'artificial intelligence' will substantially be solved".<ref>{{Harvtxt|Minsky|1967|p=2}} quoted in {{Harvtxt|Crevier|1993|p=109}}</ref> They had, however, underestimated the difficulty of the problem.{{Efn|[[Stuart J. Russell|Russell]] and [[Norvig]] write: "in almost all cases, these early systems failed on more difficult problems"{{Sfnp|Russell|Norvig|2021|p=21}}}} In 1974, both the U.S. and British governments cut off exploratory research in response to the [[Lighthill report|criticism]] of [[Sir James Lighthill]]{{Sfnp|Lighthill|1973}} and ongoing pressure from the U.S. Congress to [[Mansfield Amendment|fund more productive projects]].{{Sfn|NRC|1999|pp=212–213}} [[Marvin Minsky|Minsky]]'s and [[Papert]]'s book ''[[Perceptron]]s'' was understood as proving that [[artificial neural networks]] would never be useful for solving real-world tasks, thus discrediting the approach altogether.{{Sfnp|Russell|Norvig|2021|p=22}} The "[[AI winter]]", a period when obtaining funding for AI projects was difficult, followed.<ref name="First AI Winter">First [[AI Winter]], [[Lighthill report]], [[Mansfield Amendment]]: {{Harvtxt|Crevier|1993|pp=115–117}}, {{Harvtxt|Russell|Norvig|2021|pp=21–22}}, {{Harvtxt|NRC|1999|pp=212–213}}, {{Harvtxt|Howe|1994}}, {{Harvtxt|Newquist|1994|pp=189–201}}</ref>

<!-- 1980s and second AI winter -->
In the early 1980s, AI research was revived by the commercial success of [[expert system]]s,<ref>[[Expert systems]]: {{Harvtxt|Russell|Norvig|2021|pp=23, 292}}, {{Harvtxt|Luger|Stubblefield|2004|pp=227–331}}, {{Harvtxt|Nilsson|1998|loc=chpt. 17.4}}, {{Harvtxt|McCorduck|2004|pp=327–335, 434–435}}, {{Harvtxt|Crevier|1993|pp=145–162, 197–203}}, {{Harvtxt|Newquist|1994|pp=155–183}}</ref> a form of AI program that simulated the knowledge and analytical skills of human experts. By 1985, the market for AI had reached over a billion dollars. At the same time, Japan's [[fifth generation computer]] project inspired the U.S. and British governments to restore funding for [[academic research]].<ref name="Fund01">Funding initiatives in the early 1980s: [[Fifth Generation Project]] (Japan), [[Alvey]] (UK), [[Microelectronics and Computer Technology Corporation]] (US), [[Strategic Computing Initiative]] (US): {{Harvtxt|McCorduck|2004|pp=426–441}}, {{Harvtxt|Crevier|1993|pp=161–162, 197–203, 211, 240}}, {{Harvtxt|Russell|Norvig|2021|p=23}}, {{Harvtxt|NRC|1999|pp=210–211}}, {{Harvtxt|Newquist|1994|pp=235–248}}</ref> However, beginning with the collapse of the [[Lisp Machine]] market in 1987, AI once again fell into disrepute, and a second, longer-lasting winter began.<ref name="Second AI Winter">Second [[AI Winter]]: {{Harvtxt|Russell|Norvig|2021|p=24}}, {{Harvtxt|McCorduck|2004|pp=430–435}}, {{Harvtxt|Crevier|1993|pp=209–210}}, {{Harvtxt|NRC|1999|pp=214–216}}, {{Harvtxt|Newquist|1994|pp=301–318}}</ref>

<!-- Embodied robotics, uncertain reasoning, and connectionism in the 1980s -->
Up to this point, most of AI's funding had gone to projects that used high-level [[symbolic AI|symbols]] to represent [[mental objects]] like plans, goals, beliefs, and known facts. In the 1980s, some researchers began to doubt that this approach would be able to imitate all the processes of human cognition, especially [[machine perception|perception]], [[robotics]], [[machine learning|learning]] and [[pattern recognition]],{{Sfnp|Russell|Norvig|2021|p=24}} and began to look into "sub-symbolic" approaches.{{Sfnp|Nilsson|1998|p=7}} [[Rodney Brooks]] rejected "representation" in general and focussed directly on engineering machines that move and survive.{{Efn|
[[embodied mind|Embodied]] approaches to AI{{Sfnp|McCorduck|2004|pp=454–462}} were championed by [[Hans Moravec]]{{Sfnp|Moravec|1988}} and [[Rodney Brooks]]{{Sfnp|Brooks|1990}} and went by many names: [[Nouvelle AI]].{{Sfnp|Brooks|1990}} [[Developmental robotics]].<ref>[[Developmental robotics]]: {{Harvtxt|Weng|McClelland|Pentland|Sporns|2001}}, {{Harvtxt|Lungarella|Metta|Pfeifer|Sandini|2003}}, {{Harvtxt|Asada|Hosoda|Kuniyoshi|Ishiguro|2009}}, {{Harvtxt|Oudeyer|2010}}</ref>
}} [[Judea Pearl]], [[Lofti Zadeh]], and others developed methods that handled incomplete and uncertain information by making reasonable guesses rather than precise logic.<ref name="Stoch"/>{{Sfnp|Russell|Norvig|2021|p=25}} But the most important development was the revival of "[[connectionism]]", including neural network research, by [[Geoffrey Hinton]] and others.<ref>{{Harvtxt|Crevier|1993|pp=214–215}}, {{Harvtxt|Russell|Norvig|2021|pp=24, 26}}</ref> In 1990, [[Yann LeCun]] successfully showed that [[convolutional neural networks]] can recognize handwritten digits, the first of many successful applications of neural networks.{{Sfnp|Russell|Norvig|2021|p=26}}

<!-- 1990s: narrow, formal AI and its detractors -->
AI gradually restored its reputation in the late 1990s and early 21st century by exploiting formal mathematical methods and by finding specific solutions to specific problems. This "[[narrow AI|narrow]]" and "formal" focus allowed researchers to produce verifiable results and collaborate with other fields (such as [[statistics]], [[economics]] and [[mathematical optimization|mathematics]]).<ref name="Formal and narrow methods adopted in the 1990s">[[#Neat vs. scruffy|Formal]] and [[#Narrow vs. general AI|narrow]] methods adopted in the 1990s: {{Harvtxt |Russell|Norvig|2021|pp=24–26}}, {{Harvtxt|McCorduck|2004|pp=486–487}}</ref> By 2000, solutions developed by AI researchers were being widely used, although in the 1990s they were rarely described as "artificial intelligence" (a tendency known as the [[AI effect]]).<ref>AI widely used in the late 1990s: {{Harvtxt|Kurzweil|2005|p=265}}, {{Harvtxt|NRC|1999|pp=216–222}}, {{Harvtxt|Newquist|1994|pp=189–201}}</ref>
However, several academic researchers became concerned that AI was no longer pursuing its original goal of creating versatile, fully intelligent machines. Beginning around 2002, they founded the subfield of [[artificial general intelligence]] (or "AGI"), which had several well-funded institutions by the 2010s.<ref name="Artificial general intelligence"/>

<!--DEEP LEARNING BOOM 2012–present-->
[[Deep learning]] began to dominate industry benchmarks in 2012 and was adopted throughout the field.<ref name="Deep learning revolution">[[Deep learning]] revolution, [[AlexNet]]: {{Harvtxt|Goldman|2022}}, {{Harvtxt|Russell|Norvig|2021|p=26}}, {{Harvtxt|McKinsey|2018}}</ref>
For many specific tasks, other methods were abandoned.{{Efn|Matteo Wong wrote in [[The Atlantic]]: "Whereas for decades, computer-science fields such as natural-language processing, computer vision, and robotics used extremely different methods, now they all use a programming method called "deep learning". As a result, their code and approaches have become more similar, and their models are easier to integrate into one another."{{Sfnp|Wong|2023}}}}
Deep learning's success was based on both hardware improvements ([[Moore's law|faster computers]],<ref>[[Moore's Law]] and AI: {{Harvtxt|Russell|Norvig|2021|pp=14, 27}}</ref> [[graphics processing unit]]s, [[cloud computing]]{{Sfnp|Clark|2015b}}) and access to [[big data|large amounts of data]]<ref>[[Big data]]: {{Harvtxt|Russell|Norvig|2021|p=26}}</ref> (including curated datasets,{{Sfnp|Clark|2015b}} such as [[ImageNet]]). Deep learning's success led to an enormous increase in interest and funding in AI.{{Efn|Jack Clark wrote in [[Bloomberg News|Bloomberg]]: "After a half-decade of quiet breakthroughs in artificial intelligence, 2015 has been a landmark year. Computers are smarter and learning faster than ever", and noted that the number of software projects that use machine learning at [[Google]] increased from a "sporadic usage" in 2012 to more than 2,700 projects in 2015.{{Sfnp|Clark|2015b}}}} The amount of machine learning research (measured by total publications) increased by 50% in the years 2015–2019.{{Sfnp|UNESCO|2021}}

<!-- ALIGNMENT PROBLEM -->
[[File:20250202 "AI" (search term) on Google Trends.svg|thumb|The number of Google searches for the term "AI" accelerated in 2022.]]
In 2016, issues of [[algorithmic fairness|fairness]] and the misuse of technology were catapulted into center stage at machine learning conferences, publications vastly increased, funding became available, and many researchers re-focussed their careers on these issues. The [[AI alignment|alignment problem]] became a serious field of academic study.{{Sfnp|Christian|2020|pp=67, 73}}

<!-- AI Boom 2020-present -->
In the late 2010s and early 2020s, AGI companies began to deliver programs that created enormous interest. In 2015, [[AlphaGo]], developed by [[DeepMind]], beat the world champion [[Go player]]. The program taught only the game's rules and developed a strategy by itself. [[GPT-3]] is a [[large language model]] that was released in 2020 by [[OpenAI]] and is capable of generating high-quality human-like text.<ref>{{Cite web |last=Sagar |first=Ram |date=2020-06-03 |title=OpenAI Releases GPT-3, The Largest Model So Far |url=https://analyticsindiamag.com/open-ai-gpt-3-language-model |url-status=live |archive-url=https://web.archive.org/web/20200804173452/https://analyticsindiamag.com/open-ai-gpt-3-language-model |archive-date=2020-08-04 |access-date=2023-03-15 |website=Analytics India Magazine}}</ref> [[ChatGPT]], launched on November 30, 2022, became the fastest-growing consumer software application in history, gaining over 100 million users in two months.<ref>{{Cite news |last=Milmo |first=Dan |date=2023-02-02 |title=ChatGPT reaches 100 million users two months after launch |url=https://www.theguardian.com/technology/2023/feb/02/chatgpt-100-million-users-open-ai-fastest-growing-app |access-date=2024-12-31 |work=The Guardian |language=en-GB |issn=0261-3077 |archive-date=3 February 2023 |archive-url=https://web.archive.org/web/20230203051356/https://www.theguardian.com/technology/2023/feb/02/chatgpt-100-million-users-open-ai-fastest-growing-app |url-status=live }}</ref> It marked what is widely regarded as AI's breakout year, bringing it into the public consciousness.<ref>{{Cite web |last=Gorichanaz |first=Tim |date=2023-11-29 |title=ChatGPT turns 1: AI chatbot's success says as much about humans as technology |url=https://theconversation.com/chatgpt-turns-1-ai-chatbots-success-says-as-much-about-humans-as-technology-218704 |access-date=2024-12-31 |website=The Conversation |language=en-US |archive-date=31 December 2024 |archive-url=https://web.archive.org/web/20241231073513/https://theconversation.com/chatgpt-turns-1-ai-chatbots-success-says-as-much-about-humans-as-technology-218704 |url-status=live }}</ref> These programs, and others, inspired an aggressive [[AI boom]], where large companies began investing billions of dollars in AI research. According to AI Impacts, about $50 billion annually was invested in "AI" around 2022 in the U.S. alone and about 20% of the new U.S. Computer Science PhD graduates have specialized in "AI".{{Sfnp|DiFeliciantonio|2023}} About 800,000 "AI"-related U.S. job openings existed in 2022.{{Sfnp|Goswami|2023}} According to PitchBook research, 22% of newly funded [[Startup company|startups]] in 2024 claimed to be AI companies.<ref>{{cite web | title=Nearly 1 in 4 new startups is an AI company | website=PitchBook | date=2024-12-24 | url=https://pitchbook.com/news/articles/nearly-1-in-4-new-startups-is-an-ai-company | access-date=2025-01-03}}</ref>