Space has long been considered the final frontier for human exploration. Now it is also becoming a testing ground for artificial intelligence.
In May, NASA and IBM deployed a compressed AI model on the International Space Station and on a satellite. It was the first of its kind to demonstrate, in space, the ability to recognize floods and clouds from orbit.
In July, astronauts on the ISS tested a large language model. It helped answer questions about maintenance procedures.
These tests reflect a shift in space engineering. For decades, engineers on Earth precisely dictated the behavior of spacecraft. Now researchers are testing whether non-deterministic systems like generative AI can give spacecraft more flexibility in interpreting their environment and planning tasks.
The technology is still far from being trusted with controlling a spacecraft. But engineers are beginning to ask: can we do without such autonomy as missions become more complex, more distant and more numerous?
Robert Ambrose, former head of NASA's software, robotics and simulation division, notes that autonomy often does not produce a deterministic result. Engineers long avoided such unpredictability, but now they are looking for ways to manage it.
Long-range missions, for example to Europa, make autonomy necessary: responding to sudden events such as water plumes is impossible without onboard decisions.
The commercial boom in space is accelerating the adoption of such technologies, although the physics of microgravity requires adapting models trained on Earth.
