In a significant shift for its selection process, Roscosmos will integrate an AI system to evaluate essays written by astronaut candidates. First Deputy Prime Minister Denis Manturov confirmed that the GigaChat model will assess the logic, textual quality, and provide an initial analysis of the psycho-emotional state of applicants. This move is part of a broader, ongoing open recruitment drive that began in February 2026.

The announcement underscores a deliberate strategy to automate high-volume screening tasks. Manturov positioned this not as AI making final decisions, but as a tool to offload routine analysis, accelerating initial candidate review. This pattern is established elsewhere in the Russian space program. He cited the 'Digital Earth' platform, where algorithms automatically monitor forestry, detect land use violations, and identify infrastructure, processing satellite imagery far faster than manual methods.

Looking forward, the technical architecture points to increased orbital computing. A neural network is already aboard the International Space Station, helping crews compile experiment reports. The planned Russian Orbital Station (ROS) envisions a more powerful on-board computational complex. This system would process raw data from scientific instruments and satellites in orbit, using neural networks to identify anomalies, model atmospheric conditions, and generate finished products like maps and reports before transmission to Earth. The engineering trade-off is clear: shifting heavy computation upstream reduces both latency and ground station workload, compressing what was once a days-long process into hours.

Source: RIA Novosti