Russia's Lag in Artificial Intelligence

"Russia must become a global leader not only in production but also in creating the conditions for the penetration of artificial intelligence into all areas of life, without exception," stated Vladimir Putin, the President of Russia, at an AI conference in Moscow organized by Sberbank in December 2024. He explicitly called for AI to play an important role in Russia. During the conference attended by guests primarily from 'friendly' countries and BRICS members, he mentioned that the language model 'Gigachat,' launched under Sberbank, was used to prepare questions for Putin's direct communications with the Russian people. Putin also referred to the threats posed by what he claimed was 'Western AI,' asserting that 'they seek to destroy Russian culture and are against our morals.' Other Russian officials have also called for the development of 'Russian AI' and the elimination of 'Western AI.' Dmitry Medvedev, the former President of Russia who was previously an advocate for relations with the West and has now become one of its fiercest critics in Russia, targeted an intelligent device named 'Alice,' produced by Yandex, Russia's largest computing company, in the spring of 2024. This device refused to answer a question about Ukrainian memorials honoring Stepan Bandera, provoking strong criticism from this high-ranking Russian official. Medvedev labeled the creator of 'Alice' as 'a cowardly incompetent' and threatened Yandex executives with being recognized as 'foreign agents,' a designation that imposes numerous restrictions and is a common tool of the Russian government to suppress dissent. Kremlin propaganda is led by whom? What was 'Alice's' crime? Stepan Bandera, a Ukrainian leader, waged guerrilla warfare against Soviet forces in the early 20th century and allied with the Nazis during World War II. He has become a symbol in Russian political propaganda that 'Ukraine is a fascist state.' 'Alice's' refusal to list Bandera's memorials in Ukraine challenged this Kremlin narrative. Essentially, Russian AI suddenly rose against Kremlin propaganda. On the other hand, following the emergence of the Chinese startup 'DeepSik,' which made waves with its new AI assistant, Russia's lack of progress in this field has become even more apparent. Experts cite political limitations, brain drain, and Western sanctions that have cut off access to critical hardware as reasons for Moscow's failures in this arena. 'A hopeful beginning' five years ago, Putin's demands seemed far from impossible. At that time, Russian engineers attracted global attention due to their speed and low costs in creating efficient AI. In 2022, Sberbank, the financial backer of AI development in Russia, introduced 'Kandinsky,' an intelligent text-to-image model designed to compete with DALL-E and Midjourney from OpenAI, positioning Russia as a powerful force in the creative AI space. Additionally, Russia had a good track record in developing artificial neural networks, which mimic the brain's learning methods and are the driving force behind AI. Russian computer engineers carried out all this work in collaboration with and support from the government. For example, a facial recognition system has been connected to Moscow's metro surveillance cameras since 2020. An AI expert closely associated with Sberbank told Radio Free Europe/Radio Liberty on the condition of anonymity, 'At that time, Russia had the fortune and ability to become a global leader in AI-generated images.' However, everything collapsed in February 2022 when Russia invaded Ukraine. Shortly after Russia's large-scale military invasion of Ukraine, the U.S. and the European Union targeted Russian advanced technology personnel with sanctions. In an effort to limit the Kremlin's military capabilities and slow down the Russian economy, the sale of semiconductors, advanced processors, and dual-use technologies to Russia was prohibited. This development had a shocking impact. The American multinational company 'NVIDIA,' which specializes in developing graphics processing unit technology and is a global leader in AI hardware and software development, immediately halted shipments of chips to Russia. The Taiwanese semiconductor company, which uses American technology, also cut off its product supply to Russia. According to an expert familiar with Sberbank's internal affairs, its companies managed to stockpile about 9,000 graphics processors over the past three years. In comparison, Microsoft purchased nearly 500,000 graphics processors in 2024. Russia has attempted to circumvent sanctions by purchasing advanced components from China and India, but Indian and Chinese companies have shown little interest in these transactions due to fears of sanctions. In this situation, Russia has been forced to turn to countries like Kazakhstan to buy the necessary equipment and technology through them, but imports from 'friendly' countries are only possible in limited volumes and are accompanied by significant delays. Furthermore, producers have no obligation regarding defective items delivered to Russia. Russia is not only facing a shortage of components but also a shortage of human resources, which has become a serious challenge since the onset of the war against Ukraine. A few months after Sberbank launched the Kandinsky model in the summer of 2022, a significant number of specialists involved in the project left Russia. An AI expert told Radio Free Europe/Radio Liberty, 'I spoke with one of the engineers from this team, and he jokingly said that after the war began, almost all the leading Russian specialists in image production with neural networks ended up in a rented apartment in Yerevan and quickly found jobs in Western companies after this stop.' Sberbank downplays the impact of brain drain on its operational projects, but statistics tell a different story. Mikhail Shadayev, Russia's Minister of Digital Development and Mass Communications, announced that around 100,000 computer engineers, constituting nearly ten percent of the country's total technical staff, left Russia in 2022; some immediately after the war began and others following the announcement of general mobilization. Russian tech specialists now reside in various countries, primarily the UAE, Georgia, Armenia, and Serbia, and often have little desire to return to Russia due to higher salaries and a more constructive environment for international collaboration. The Russian government has attempted to stop the brain drain by increasing salaries and establishing new research centers in Kazan, Novosibirsk, and other cities, but these efforts have been merely a drop in the ocean. In addition to technical and human challenges, ideological constraints pose another barrier to advanced technologies in Russia. Echoing the criticisms of Putin and Medvedev, Alexander Dugin, a theorist close to the Kremlin, called in December 2024 for the creation of a version of AI that would 'immediately confirm not only Moscow's dominance over Crimea but also the belonging of Kyiv, Kharkiv, and Odessa to Russia.' By the end of 2024, Putin ordered the Russian government and Sberbank to expand cooperation with China in the field of AI. Reuters reported that Sberbank intends to collaborate with Chinese researchers on joint AI-related projects. Since 2023, China has been censoring large language models. This censorship occurs not only at the final tuning stage, which prevents AI from answering certain questions, but also at the training stage, where this AI model does not have access to pre-selected data. With the limitations faced by AI specialists in Russia, the establishment of a stringent censorship system, which is another blow to freedom of expression, exerts excessive pressure. China is able to do this because its AI industry is extensive and, unlike Russia, has vast datasets, advanced infrastructure, and deep connections with other producers worldwide. At least as long as international sanctions remain in place, it is unlikely that Russian AI will accelerate beyond its current slow progress. In the 'Chatbot Arena' ranking, Russian large language models produced by Sberbank and Yandex occupy the 12th and 17th positions among 44 models. The top three models in this list are: Gemini Pro 1.5 from Google, Claude 3.5 from Anthropic, and ChatGPT-4 from OpenAI. The negative impact of falling behind in AI development is not limited to chatbots and intelligent language models. Given the increasing role of AI in scientific research, industrial innovations, and military capabilities, Russia may lose its economic competitiveness, technological sovereignty, and strategic influence on a global scale.