# Generative AI experiments point to more autonomous spacecraft, IEEE Spectrum reports

- Published: 2026-09-28T11:00:05.000Z
- Updated: 2026-09-28T14:28:23.242Z
- Section: AI in practice
- Topics: Research, Agents
- Page: https://aidailyjournal.com/en/n/2026/09/28/generative-ai-experiments-point-to-more-autonomous-spacecraft-ieee-spectrum-repo/
- Sources: [IEEE Spectrum](https://spectrum.ieee.org/generative-ai-in-space-exploration)

## Summary

IEEE Spectrum reports that NASA’s Jet Propulsion Laboratory used Anthropic’s Claude models to help plan two drives by the Perseverance rover on Mars, with humans reviewing the route before upload. In May, NASA and IBM placed a compressed AI model on the International Space Station (ISS) and a satellite to identify phenomena such as floods and clouds; in July, astronauts tested a large language model on questions about maintenance procedures. The magazine describes these experiments as part of a shift toward systems that can interpret their surroundings and plan tasks, although they are not yet ready to take control of a spacecraft.

Robert Ambrose, a former head of a NASA software, robotics and simulation division, says nondeterministic outcomes make testing difficult, but argues that autonomy could be essential for distant missions such as a possible trip to Europa. Icarus Robotics also plans to send its Joy robot to the ISS, starting with teleoperation and using the data to develop autonomy.

## Why it matters (editorial interpretation)

The experiments described connect generative models to operational tasks in space, but human review and the difficulty of testing nondeterministic outputs still limit autonomy. They help illustrate the gap between demonstrating usefulness in missions and allowing a spacecraft to take control.

_Summary written by AI Daily Journal in its own words, based on the sources above. For details, read the original articles._