Andrey Zubarev, Senior Researcher of the Department for Digital Finance at the Gaidar Institute, commented for RTVI on the role of Artificial intelligence in the military sphere.
He believes, AI no longer simply assists in intelligence, data processing, and operational planning; it determines which decision options are accepted for consideration by Command. However, it is too early to name it a fully-fledged infrastructure, as technologies remain fragmented and unreliable.
According to the expert, the main risk is not associated with autonomous drones, rather with decision-making systems: an error there could result not in a single failure, but in a collapse of an entire strategy. Another threat is dependence on external suppliers: more than 80% of the AI chip market is controlled by a single manufacturer, creating thereby structural vulnerabilities.
Andrey Zubarev identified three groups of countries. The US and China possess a full technology stack, from chips to ready-made solutions. Allies (the EU and Japan) partially depend on partners. The remaining countries use ready-made technologies without controlling key components. The most important element for sovereignty is computing infrastructure and specialized software (integrators). Even having their own chips doesn't provide complete control without such software.
According to Andrey Zubarev, Russia faces a dual situation. On the one hand, there is an infrastructure shortage and dependence on foreign chips. On the other, there is a legal requirement to use only domestically produced "trusted models" in critical systems. This limits both imports and the development of domestic solutions, creating the risk of systemic lag in the military application of AI.