Vladimir Putin has warned that Russia will develop its own advanced AI models to counter Western influence over the technology – amid warnings that such tools could be misused by nations to influence elections or galvanise support for wars.
The Russian president has so far blamed the West for his force’s failure to overwhelm Kyiv, setting Moscow on course for a new digital arms race against the US and UK for AI supremacy.
Putin has claimed that AI models like OpenAI’s ChatGPT and Google’s Bard chatbots “cancel Russian culture” and that the West holds a “dangerous” dominance of the technology
“Our innovations should rest on our traditional values [and] the wealth and beauty of the Russian language,” Mr Putin added, in language similar to a number of other recent speeches railing against the West.
Geoffrey Hinton, a British-Canadian computer scientist often dubbed the “godfather of AI”, referred to the Russian president in an interview earlier this year saying that Mr Putin or others may “want to use them for winning wars or manipulating electorates”.
The Russian president was addressing an AI conference in Moscow on Friday where he said Russian investment in AI development was being increased across all sectors.
Citing the example of Gazprom Neft, Mr Putin said one of Russia’s largest oil producers was using AI to slash the cost of oil well development and to address complicated logistics safety issues.
“I hope we will be more active in this area. When I say ‘we,’ I am referring not only to the government but also to the regions and industries, and individual plants,” Mr Putin said.
The Russian leader said the country would intensify its research into the domains of generative AI and large language models which currently lag behind the leading Western-developed tools like OpenAI’s ChatGPT and Google’s Bard chatbots.
To achieve such development, he called for the scaling up of Russia’s supercomputing power and for improvements to its top-level AI education programmes.
“Our domestic models of artificial intelligence must reflect the entire wealth and diversity of world culture, the heritage, knowledge, and wisdom of all civilisations,” he said.
English-speaking countries currently dominate AI development, with Stanford’s Institute for Human-Centered Artificial Intelligence (HAI) claiming the US and UK were further ahead in the technology than the rest of the world.
A growing number of studies are revealing the bias in AI models developed in primarily English-speaking countries.
Another bias: If you're a non-native English writer, GPT detectors will frequently classify your writing as #AI-generatedhttps://t.co/xPAeCstivF @james_y_zou @liang_weixin @Stanford pic.twitter.com/0qUygrEuac
— Eric Topol (@EricTopol) July 10, 2023
For instance, a study published in July in the journal Cell, showed that GPT detectors frequently classify actual writing by non-native English speakers as AI-generated.
Observers have also noted that OpenAI’s latest GPT-4 AI model excels in English, Spanish, Italian, Indonesian, and other Latin alphabet-based languages, while it struggles with Thai, Punjabi, and other languages with different alphabets.
Even Facebook parent company Meta warned in July that since most of the training data used for its updated AI large language model is in English, it “may not be suitable for use in other languages.”