Artificial intelligence installed on the ISS

The US company Booz Allen Hamilton, whose specialization is related to intelligence and counterintelligence in the field of electronic services, announced the successful deployment of generative artificial intelligence on the International Space Station. The AI ​​was launched on an HPE server, the updated configuration of which was delivered to the ISS in January 2024. The AI ​​at the station will assist the crew and provide primary processing of the data collected by the station.

HPE Spaceborne Computer-2. Image source: NASA

Booz Allen Hamilton Large Language Model (LLM) runs on HPE Edgeline EL4000 single-GPU x86 servers and HPE DL360 Gen10. The platform received an impressive SSD block from KIOXIA with a capacity of 130 TB compared to previous times. This will help collect and process more data on board the ISS, which promises to reduce the amount of data transmitted to Earth by up to 30 thousand times. The ISS computer will send mostly finished analytics and other final results, rather than raw raw material, as before.

«Generative AI in space is truly a new frontier, and this opportunity unlocks the potential of generative AI in orbit to integrate and develop mission-critical solutions and aligns with Booz Allen Hamilton’s mission to create human-driven AI solutions that enable our nation to thrive.” – said the company, widely known as the employer of Edward Stone (Edward Snowden).

The company says the model, developed in just over eight weeks, will help frontline crews efficiently obtain relevant information, understand missions, and solve complex problems using natural language processing. In the future, such experience will help autonomous settlements in the depths of the solar system and beyond, but it all begins here and now.