AI before AI: The forgotten Soviet programme (1/2)
France 24 · LC · trust 52/100
Beginning in the 1950s, the Soviet Union launched an ambitious programme aimed at developing artificial intelligence. While the history of AI today seems to be shaped by the giants of Silicon Valley, ideas originating from behind the Iron Curtain had a significant influence, although they remain largely overlooked.
By: Sébastian SEIBT The Soviet Union took an interest in artificial intelligence as early as the 1950s. © Studio graphique France Médias Monde American chess genius Bobby Fischer defeated the reigning world champion Russia 's Boris Spassky in 1972 in what has been called "the most exciting world chess championship ever".
But the Eastern Bloc got its revenge on August 8, 1974, when Kaissa, a Soviet computer, checkmated its opponent, securing the very first World Computer Chess Championship title for the Soviet Union . Russian artificial intelligence had crushed its capitalist competitor.
The victory proved to the world that the Soviets also had a say in the field of AI, which had until then mostly been developed at top US universities.
In the early 1950s, AI – which was not yet known by that name – was viewed behind the Iron Curtain with suspicion. The Soviets viewed what was to become AI primarily through the lens of cybernetics, a field of research that was very much in vogue for studying human-to-machine or machine-to-machine behaviour.
Some saw cybernetics as a “false bourgeois science”, while others thought it held great promise for scientific advancement.
“Cybernetics almost became a sort of official philosophy of scientific research in the USSR,” said Olessia Kirtchik, a sociologist at the European Centre for Sociology and Political Science who has worked on the history of AI in the Soviet era .
The hallways of research institutes and universities were then filled with mathematicians, philosophers and other researchers, tasked with ensuring the communist regime reaped the benefits of cybernetics. While it fell out of favour in the US by the late 1950s, giving way to computer science and what eventually became AI, the Soviets remained committed to cybernetics, particularly the concept of the “thinking machine”.
One of the leading figures in Soviet AI to go down in history was Alexander Kronrod , a mathematician who was, among other things, in charge of the team that developed the Kaissa chess programme. For some, Kronrod was even the Soviet “father of artificial intelligence”.
But before him, another mathematician had a considerable influence on the development of AI. His name was Dmitry Pospelov.
Pospelov brought prestige to the discipline by “fostering and supporting the artificial intelligence research community for years”, said Kirtchik. This “lobbying” work culminated in the creation of the Russian Association for Artificial Intelligence in 1989.
Pospelov’s pro-AI campaign within Moscow’s corridors of power also illustrates how the field was not immune to the ideological battle of the Cold War era. The Russian mathematician strongly criticised the Western approach to AI, which he deemed too “reductionist”. In his view, American researchers were reducing the human brain to a sort of supercomputer, performing rigorous but disembodied calculations. For him, intelligence had to be conceived as a social activity dependent on its environment, rather than as a purely logical process.
In the Soviet Union, AI was expected to solve practical problems, and the USSR primarily “used algorithms to optimise the management of the socialist economy”, according to Kirtchik.
The oil and gas industries, for example, used it to improve their operations.
To display this content from YouTube, you must enable advertisement tracking and audience measurement.
One of your browser extensions seems to be blocking the video player from loading. To watch this content, you may need to disable it on this site.
This may have been one of the first practical success of “machine learning”, the ability of an algorithm to learn from the data it is given. Others see it as proof that AI does not necessarily need huge datasets to find the right answers.
“With clever methods, small data can move mountains – or in this case, reveal them,” Valery Manokhin, a machine learning specialist, wrote in a blog post on Medium on the Soviet gold rush.
Some of Silicon Valley’s big names also have Soviet AI to thank. Apple might never have released its Newton tablet in 1993 without the work of Shelia Guberman, the Soviet mathematician who solved a puzzle that had been troubling many Western AI specialists: how to get machines to recognise handwriting.
“He achieved this by adopting an approach known as Gestalt (a form of pattern recognition), which had been overlooked in the West,” Kirtchik said.
Guberman's solution was adopted by Soviet entrepreneur Stepan Pachikov for the Paragraph company. His handwriting recognition software was used by Apple to develop the Newton personal digital assistant; the technology was subsequently sold to Microsoft and used by the US Postal Service.
Other ideas that were developed in the Soviet Union are now enjoying a resurgence, including the “automaton” conceived by Michael Tsetlin in the early 1960s. These ideas inspired Norwegian computer scientist Ole-Christoffer Granmo, who in 2024 introduced the idea of the “ Tsetlin machine ”, designed to be a more energy-efficient alternative to today’s large language models like ChatGPT .
But Tsetlin's ideas would likely annoy tech bros of today like Sam Altman and Elon Musk , however. For Tsetlin, “the workings of these automata had to be perfectly transparent, unlike the black boxes that are today’s AI models”, Kirtchik said.
The Soviets were not short of ideas, but as was often the case during the Cold War , they simply did not have the same resources as the West.
The practical applications of Soviet research often remained “marginal", Kirtchik said, "since the computers at their disposal were often incapable of performing…
Read the original at France 24 →
Open in TruthVane →