Planetary scientests from Nasa’s Jet Propulsion Laboratory developed the machine learning tool to save researchers time, training the AI with 6,830 images of the Red Planet.
It has already found a crater that was formed between March 2010 and May 2012.
This crater was very small compared to others – only four meters in diameter – making it difficult for human scientists to detect.
The tool that found it, called an automated fresh impact crater classifier, used pictures taken from Nasa’s Mars Reconaissance Orbiter (MRO).
It looks for events like dust devils, avalances, and shifting dunes, and has been used to find over 1,000 craters.
However, images from the MRO can usually only pick up the blast marks around a crater’s impact, rather than the crater itself.
Without artificial intelligence, scientists then need to examine those pictures the High-Resolution Imaging Science Experiment (HiRISE).
The task is time-consuming; it can take a researcher 40 minutes to scan an image properly.
This new artificial intelligence, once it was trained, was set to analyse the full library of 112,000 images taken by the Context Camera.
Running on a supercomputer, the AI is capable of detecting craters at a speed 480 times faster than humans – cutting the 40 minute detection time down to only five seconds.
750 copies of the classifier were ran simultaneously. “It wouldn't be possible to process over 112,000 images in a reasonable amount of time without distributing the work across many computers,” said JPL computer scientist Gary Doran.
“The strategy is to split the problem into smaller pieces that can be solved in parallel.”
Despite the achievements of the artificial intelligence, it still requires a human being to check its work for accuracy because of its inability to do more skilled analysis.
“Tools like this new algorithm can be their assistants. This paves the way for an exciting symbiosis of human and AI 'investigators' working together to accelerate scientific discovery,” said JPL computer scientist Kiri Wagstaff.
Eventually, the aim is for systems like these to run on computers onboard Mars orbiters, rather than being ran on computers on Earth.
Currently, the data being sent back to Earth still requires scientists to examine it – a task Michael Munja, a Georgia Tech graduate student who worked on the classifier, compared to finding a needle in a haystack.
“The hope is that in the future, AI could prioritize orbital imagery that scientists are more likely to be interested in,” Munje said.
It is also hoped that the tool could offer a more complete vision of how often meteors strike mars, as well as finding even smaller impacts that have been overlooked by scientists.
“There are likely many more impacts that we haven't found yet,” scientist Ingrid Daubar said.
“This advance shows you just how much you can do with veteran missions like MRO using modern analysis techniques.”
This is not the only instance where artificial intelligence has been used by Nasa to detect data missed by human scientists.
An artificial intelligence algorithm has discovered 50 new potential planets that were missed by humans.
The 50 planets range from the size of Neptune to smaller than the Earth. Some had orbits that last as long as 200 days on Earth, while others spin around their respective stars as quickly as once a day.
Another AI tool has been developed to reveal the structure of the universe, building a tool called the "Dark Emulator" that can create hundreds of virtual universes and use those simulations to help scientists find out more about our reality.