Google AI helped find the first solar system outside our own with 8 planets

Next Story

The FCC officially votes to kill net neutrality

Google and NASA today announced the discovery of a new planet in solar system Kepler 90, achieved using machine learning. By applying neural networking to Kepler data, scientists have found, for the first time, an eighth planet in the Kepler-90 system – this ties the Kepler-90 system with our own system for the most planets in any known system.

Kepler-90 i is the new planet discovered by Google’s neural network, and it had a weaker signal than the ones normally used to identify planets by traditional means. Google’s tool also found Kepler-80 g, a planet that’s the smallest in its own system.

Google explained on a call announcing the discovery that it essentially used the same tools it’s employed to do things like identify cats and dogs in photographs to comb through the data collected by Kepler over its four year data collection mission.

Kepler-90 is slightly larger, hotter and has more mass than the sun, but otherwise resembles our own star in many ways. Kepler-90 i is the smallest of the planets orbiting the star, and likely isn’t hospitable to life – it has a rocky surface, and a surface temperature of around 800 degrees Fahrenheit, NASA says. But it’s also a crucial additional signal that backs up a theory that planets closer into their stars tend to be smaller and rockier than those further out. It also rotates Kepler 90 in just 14 days, vs. the 365 day cycle for Earth around the sun.

Kepler-90 could also play host to more planets other than the ones we know about – Kepler has only searched for planets around the star in an area very near in to the solar body – NASA says that there is a strong possibility that there are more planets further out that haven’t yet been observed.

Google’s AI only examined 670 stars out of the 200,000 or so observed by Kepler in order to achieve these findings, with two brand new exoplanets revealed. That means the potential for further discoveries in the larger data set is very high.