Triff/Shutterstock
Artificial intelligence has been advancing at an exponential rate over the past few years, prompting widespread concern that AI could start taking jobs away from real people. Unfortunately for the human workforce, these fears are already coming true, even for one of the most specialized jobs in the world: NASA rover driver.Â
Rovers have been at the heart of Mars exploration efforts over the past three decades, and up to this point, they have always relied on human operators to control their movements. However, NASA closed out 2025 with an unprecedented experiment, using AI vision-language models to map and direct the Perseverance rover on a successful journey across the Martian surface. It was only a trial run for the new technology, but the resounding success makes it all but certain that rover drivers are destined to become obsolete.
NASA has landed five different rovers on Mars to date, although only the two most recent ones — Curiosity and Perseverance — are currently active. While each new rover has carried more advanced technology than the last, the basic system of operating them has remained fundamentally the same. Human operators working at NASA facilities like the Jet Propulsion Laboratory (JPL) in southern California program routes for the rover to take and send the instructions remotely to Mars. These efforts have brought us the clearest images of Mars ever captured, but they come at the expense of a lot of time and money. NASA officials are hoping that a shift to AI operation can save on both fronts.
The technical challenges of driving a Mars rover
Merlin74/Shutterstock
The distance from Earth to Mars is roughly 140 million miles, depending on where each planet is in its orbital path. At that distance, it takes around 25 minutes for a signal to pass from a NASA laboratory to one of its Martian rovers and back. This delay means that operators aren’t able to drive a rover in real time like an RC car, and instead, have to pre-program directions for the vehicle. They do this by first making digital sketches of the route, from which they create a series of directions for the rover to follow. These directions are put in a computer file and transmitted from radio dishes on Earth to each rover’s antenna.
Every route that a Mars rover takes requires NASA to program hundreds or even thousands of individual directions, and the vehicles move at a painfully slow rate. Waypoints on the route can be no more than 330 feet apart, forcing the rovers to explore in small, slow steps. In its first four years on Mars, Perseverance drove a grand total of … 25 miles. One aim of using AI to map rover routes is to generate waypoints at a faster rate, creating whole routes in less time. That could potentially help the rovers pick up their pace and survey greater reaches of the red planet.
NASA used a familiar AI system for help piloting its rover
Mijansk786/Shutterstock
You might think that NASA had to develop some hyper-advanced Mars-mapping artificial intelligence program specific to this task, but they actually used the popular Claude by Anthropic. The team at JPL fed Claude’s AI models satellite images of Mars gathered by NASA’s Mars Reconnaissance Orbiter. From this map data, the AI was given a point A and point B and tasked with creating a route between them that avoided obstacles and hazards. The AI plotted out a series of waypoints, and impressively managed to identify and plan around boulder fields and sand traps.
NASA conducted the first AI-directed rover mission on December 8, 2025, successfully piloting Perseverance across 689 feet of the Martian surface. Two days later, the rover covered another 807 feet on an AI-generated route. These were only short trial runs, but the successful results are encouraging for a grander vision of the future. The aim is to ultimately harness AI to map routes not on the scale of a few hundred feet, but on the scale of miles at a time.
It can’t be ignored that expanding AI use at NASA will come at the cost of jobs for real workers amidst a contentious political back-and-forth over the space administration’s budget. However, humans won’t be entirely out of the picture if NASA elects to assign rover-driving duties to AI. Skilled people will still be needed to check every AI-generated route and ensure that the computer doesn’t accidentally crash poor Perseverance.