Scientists train artificial intelligence to find life on Mars

Kimberley Warren-Rhodes, Senior Investigator at the SETI Institute in Mountain View, who has been searching for signs of extraterrestrial life since the early 2000s, decided to get together. their expertise in statistical ecology with emerging technologies such as artificial intelligence (AI) and machine learning. The aim is to create a comprehensive database of different scenarios similar to the planet Mars.

Naturally, the new vehicle must first be “trained” here on Earth before it can be used effectively on other planets. A team of researchers has successfully tested an AI to map any trait that could provide evidence of past or present life, “however hidden or rare they may be.”

Working in Chile’s Atacama Desert, one of the world’s most uninhabitable places, artificial intelligence has significantly reduced the three square kilometers it needs to explore. According to Warren-Rhodes, the search area was reduced by up to 97% and the probability of finding life increased by up to 88%.

How was AI trained in the Atacama Desert?

According to the article published in the journal on Monday (6) Nature BiologySince 2016, the Warren-Rhodes crew has been traveling to the nearly uninhabitable plateau of the Atacama Desert, which is considered a “generic” version of the planet Mars. At an altitude of 3,500 meters in the Chilean Andes, the researchers looked for photosynthetic organisms (which use solar energy to produce food) found in rocks called endoliths.

The authors collected everything from drone images to geochemical analyzes and DNA sequences to fully map the environment. This dataset is similar to those currently collected on Mars by orbiting satellites, drones, and rovers. What was discovered at Atacama was that endoliths were mostly found in a relatively soft, porous, water-retaining mineral called alabaster.

In a press release posted on the SETI Institute’s website, the team theorizes that algorithms similar to the experiment could be adapted to different types of potentially habitable environments in the near future. In this case, biosignatures can be automated into planetary robots that will become the “eyes” of mission planners for areas of any scale most likely to contain life.