AI winter
In plain wordsA slump in which disappointment over AI's unmet promises leads governments and companies to pull back money and attention. Historians count two big ones: one in the 1970s and one starting in the late 1980s.
The bust phase of a boom-and-bust cycle in artificial intelligence: a span of years in which money, commercial backing and scientific standing drain away from the field because earlier expectations have gone unmet. Two such periods are usually recognized. The first, roughly 1974–1980, came after critical official reviews — the ALPAC reportALPAC report, prop. n.: A 1966 US government report that concluded machine translation was slower, worse and costlier than human translators. It ended most American funding for the field for many years. (1966) on machine translation and Britain's Lighthill reportLighthill report, prop. n.: A 1973 review written for a British government research council that judged AI research to have fallen far short of its promises. It led to deep cuts in AI funding in the United Kingdom. (1973) — together with US defense agencies' retreat from open-ended research grants and a slump in neural-network work often blamed on PerceptronsPerceptrons (book), prop. n.: A 1969 book by Marvin Minsky and Seymour Papert proving that simple one-layer neural networks cannot learn certain patterns. It is widely blamed for a decade-long slump in neural-network research.. The second, from about 1987 to the mid-1990s, began as the market for Lisp machinesLisp machine, n.: A 1970s–1980s personal workstation built specially to run the AI programming language Lisp quickly. When ordinary computers caught up, the market collapsed, helping to bring on an AI winter. collapsed and expert systemsexpert system, n.: A program that stores a human specialist's know-how as many if–then rules and uses them to give advice or make decisions in a narrow field, such as diagnosing infections or configuring computers. proved hard to scale and costly to maintain, and it deepened as Japan's Fifth Generation Computer Systems projectFifth Generation Computer Systems project, prop. n.: Japan's government-backed effort, from 1982 to 1992, to build a new kind of computer that reasons with logic using many processors at once. It fell short, but it alarmed other countries into launching AI programs of their own. and DARPA's Strategic Computing InitiativeStrategic Computing Initiative, prop. n.: A billion-dollar US defense research program, run from 1983 to 1993, that tried to build machine intelligence for military uses such as self-driving vehicles and assistants for fighter pilots. wound down. During winters, researchers often relabeled their work as informatics, knowledge-based systems or machine learningmachine learning, n.: A way of building software that learns from examples instead of following rules written by a programmer. Shown thousands of labeled photos, for instance, a program can learn to recognize cats. to avoid the stigma of 'AI'.
Skeptics ask whether today's investment boom will end in a third AI winter.
Origin
Modeled on 'nuclear winter'. Its first known public use was as the theme of a 1984 panel debate at the annual meeting of the American Association for Artificial Intelligence (now the Association for the Advancement of Artificial IntelligenceAssociation for the Advancement of Artificial Intelligence, prop. n.: A major professional society for artificial intelligence researchers, founded in the United States in 1979, which runs a large yearly AI conference, publishes a magazine and honors leading scientists.). There Roger Schank and Marvin MinskyMarvin Minsky, prop. n.: An American AI pioneer who helped launch the field and MIT's AI laboratory, co-wrote a famous critique of early neural networks, and pictured the mind as a society of many simple parts., veterans of the 1970s cuts, cautioned that business excitement about expert systems had run ahead of results and predicted a crash.
Usage note
The dates, and even the number, of winters are debated; research never stopped, and some historians describe several smaller downturns. Periods of rapid growth are sometimes called AI springs or summers. See also AI bubbleAI bubble, n.: The worry that investors are pouring far more money into AI companies, chips and data centers than AI will earn back soon, so that prices could crash as in the dot-com bust of 2000. and hype cyclehype cycle, n.: A pattern in which excitement about a new technology rises to a peak, crashes into disappointment, then slowly recovers as the technology finds real uses..
Related
- Lighthill reportLighthill report, prop. n.: A 1973 review written for a British government research council that judged AI research to have fallen far short of its promises. It led to deep cuts in AI funding in the United Kingdom.
- ALPAC reportALPAC report, prop. n.: A 1966 US government report that concluded machine translation was slower, worse and costlier than human translators. It ended most American funding for the field for many years.
- Lisp machineLisp machine, n.: A 1970s–1980s personal workstation built specially to run the AI programming language Lisp quickly. When ordinary computers caught up, the market collapsed, helping to bring on an AI winter.
- expert systemexpert system, n.: A program that stores a human specialist's know-how as many if–then rules and uses them to give advice or make decisions in a narrow field, such as diagnosing infections or configuring computers.
- AI hypeAI hype, n.: Exaggerated claims about what AI can do or will soon do, made to attract money, attention or customers, which can mislead the public and policymakers.
- AI bubbleAI bubble, n.: The worry that investors are pouring far more money into AI companies, chips and data centers than AI will earn back soon, so that prices could crash as in the dot-com bust of 2000.
- hype cyclehype cycle, n.: A pattern in which excitement about a new technology rises to a peak, crashes into disappointment, then slowly recovers as the technology finds real uses.
Sources
- Crevier, D. (1993). AI: The Tumultuous History of the Search for Artificial Intelligence. New York: Basic Books.
Mentioned in
AI bubbleAI bubble, n.: The worry that investors are pouring far more money into AI companies, chips and data centers than AI will earn back soon, so that prices could crash as in the dot-com bust of 2000., AI hypeAI hype, n.: Exaggerated claims about what AI can do or will soon do, made to attract money, attention or customers, which can mislead the public and policymakers., ChatGPT momentChatGPT moment, n.: A phrase for the point when a technology suddenly becomes impressive and widely used, as ChatGPT did in late 2022; people ask when robotics, biology or other fields will get theirs., combinatorial explosioncombinatorial explosion, n.: The way the number of possible combinations — moves, routes, schedules — grows so quickly as a problem gets bigger that checking them all becomes impossible, even for the fastest computers., DARPADARPA, prop. n.: The US military's research-funding agency, which paid for much of the early development of artificial intelligence and the internet and runs prize competitions such as driverless-car races., Dartmouth workshopDartmouth workshop, prop. n.: A summer meeting of scientists at Dartmouth College in 1956, planned in a 1955 funding proposal that first used the phrase 'artificial intelligence'. Most histories treat it as the moment AI became a research field of its own., expert systemexpert system, n.: A program that stores a human specialist's know-how as many if–then rules and uses them to give advice or make decisions in a narrow field, such as diagnosing infections or configuring computers., good old-fashioned AIgood old-fashioned AI, n.: A nickname for the classic style of AI, dominant from the 1950s to the 1980s, that tried to make computers intelligent by writing knowledge and rules down by hand., hype cyclehype cycle, n.: A pattern in which excitement about a new technology rises to a peak, crashes into disappointment, then slowly recovers as the technology finds real uses., LispLisp, prop. n.: A programming language invented in 1958 for AI research, in which programs and data are written the same way, as nested lists in parentheses. It was the main language of AI for decades., Perceptrons (book)Perceptrons (book), prop. n.: A 1969 book by Marvin Minsky and Seymour Papert proving that simple one-layer neural networks cannot learn certain patterns. It is widely blamed for a decade-long slump in neural-network research., scaling wallscaling wall, n.: The disputed idea that making AI models bigger is running out of steam, so progress will soon stall; 'hitting a wall' became the skeptics' slogan and 'there is no wall' the reply., XOR problemXOR problem, n.: The classic puzzle of teaching a single artificial neuron the 'one or the other but not both' rule. It cannot be done with one straight dividing line, which showed why networks need hidden layers., ALPAC reportALPAC report, prop. n.: A 1966 US government report that concluded machine translation was slower, worse and costlier than human translators. It ended most American funding for the field for many years., Fifth Generation Computer Systems projectFifth Generation Computer Systems project, prop. n.: Japan's government-backed effort, from 1982 to 1992, to build a new kind of computer that reasons with logic using many processors at once. It fell short, but it alarmed other countries into launching AI programs of their own., Georgetown–IBM experimentGeorgetown–IBM experiment, prop. n.: A 1954 public demonstration in New York in which an IBM computer translated about sixty Russian sentences into English. It raised hopes for automatic translation that took decades to fulfill., it's over, we're so backit's over, we're so back, phr.: A meme describing the mood swings of the AI world: one day bad news means 'it's over', the next a new model or result means 'we're so back'., knowledge acquisition bottleneckknowledge acquisition bottleneck, n.: The problem that getting knowledge out of human experts and into a computer by hand is so slow and costly that it limited how far classic expert systems could go., Lighthill reportLighthill report, prop. n.: A 1973 review written for a British government research council that judged AI research to have fallen far short of its promises. It led to deep cuts in AI funding in the United Kingdom., Lisp machineLisp machine, n.: A 1970s–1980s personal workstation built specially to run the AI programming language Lisp quickly. When ordinary computers caught up, the market collapsed, helping to bring on an AI winter., microworldmicroworld, n.: A small, deliberately simplified artificial setting, such as a table of toy blocks, in which an AI program can be tested in full detail before anyone tries it in the messy real world., Strategic Computing InitiativeStrategic Computing Initiative, prop. n.: A billion-dollar US defense research program, run from 1983 to 1993, that tried to build machine intelligence for military uses such as self-driving vehicles and assistants for fighter pilots., XCONXCON, prop. n.: A rule-based program used from 1980 by the computer maker Digital Equipment Corporation to work out which parts each customer's computer order needed. It was the first big commercial success of expert systems.