Even on Mars, AI is proving it’s here to stay—and it’s taking on tasks we never imagined. Boldly, NASA’s Perseverance rover has just completed its first-ever drives on the Martian surface, not steered by human hands, but planned entirely by artificial intelligence. Yes, you read that right. Last month, Perseverance navigated not one, but two routes across the Red Planet’s treacherous terrain, all thanks to AI. And here’s the kicker: it survived to tell the tale. But here’s where it gets controversial—is this the beginning of AI replacing even the most specialized human jobs, like Mars Rover Driver? Let’s dive in.
According to NASA (https://www.nasa.gov/missions/mars-2020-perseverance/perseverance-rover/nasas-perseverance-rover-completes-first-ai-planned-drive-on-mars/), the Perseverance rover executed these drives on December 8 and 10, guided by generative AI that mapped out waypoints—a task typically handled by human mission planners. This isn’t about autonomous driving, which Perseverance already excels at (https://www.jalopnik.com/nasas-record-breaking-mars-rover-is-way-better-at-self-1848779123/). Instead, it’s about the intricate route planning that ensures the $2.7 billion rover (https://www.planetary.org/articles/cost-of-perseverance-in-context) doesn’t crash into boulders, flip on uneven ground, or get stuck in sand traps like its predecessor, the Spirit rover (https://www.jalopnik.com/this-day-in-history-spirit-exploration-rover-lands-on-1845980402/). After all, NASA’s budget can’t afford such mishaps (https://www.jalopnik.com/2082419/congress-saves-nasa-trump-proposed-budget-cuts/).
And this is the part most people miss: driving the rover in real time from Earth is impossible due to the vast distance between our planets. Even at light speed, it takes 4 to 24 minutes (https://blogs.esa.int/mex/2012/08/05/time-delay-between-mars-and-earth/) for signals to travel between Earth and Mars. Traditionally, human planners meticulously study Martian terrain data to plot safe routes, upload them to the rover, and hope for the best. But this time, NASA’s Jet Propulsion Laboratory (JPL) handed the reins to Anthropic’s Claude AI, which planned the routes after analyzing the data.
Where we’re going, we don’t need roads—or human drivers. But before you panic about AI sending Perseverance off a cliff, JPL took precautions. Remember, AI models have been known to make questionable decisions, like launching nukes in war simulations (https://www.jalopnik.com/ai-models-are-eager-to-launch-nukes-in-war-simulations-1851235624/). To avoid disaster, JPL used a ‘digital twin’ of the rover—a virtual replica—to test the AI-planned routes across 500,000 variables. Only after thorough testing did they greenlight the mission.
On December 8, Perseverance successfully navigated a 689-foot route, followed by an 807-foot journey on December 10. While the rover’s autonomous systems made minor adjustments based on real-time data, the AI-planned routes worked flawlessly. As Matt Wallace, manager of JPL’s Exploration Systems Office, put it: ‘Imagine intelligent systems not only on Earth but also in edge applications like rovers, helicopters, and drones, trained with the collective wisdom of NASA engineers, scientists, and astronauts. That’s the game-changing technology we need for a permanent human presence on the Moon and beyond.’
But let’s pause for a moment. Human astronauts drove rovers on the Moon (https://www.jalopnik.com/heres-how-to-drive-the-lunar-rover-1702630957/) long before AI was a household name. And unlike Earth’s robotaxis, Martian rovers don’t have to worry about felony stops (https://www.jalopnik.com/2042619/waymo-robotaxi-cruises-through-felony-stop/), stopped school buses (https://www.jalopnik.com/2053574/waymo-fix-robotaxis-school-buses/), or children darting into the street (https://www.jalopnik.com/2091500/waymo-hits-child-near-elementary-school/)—unless there are alien kids we haven’t met yet. While we joke about AI taking over route planning jobs, it makes practical sense for a rover to scan its surroundings, combine that data with existing maps, and plan its own route without waiting for instructions from Earth. As JPL notes, it would speed up exploration significantly.
But here’s the question: Are we ready to trust AI with tasks as critical as this? And what does it mean for the future of human expertise in space exploration? Let us know your thoughts in the comments—do you think AI should take the wheel, or should humans remain in control? The debate is just beginning.
