A team of researchers from the University of California, Berkeley, has developed an Artificial Intelligence (AI) that uses deep learning search to look for extraterrestrial life in 820 nearby stars and, so far, the results are promising.
What Does Detecting Signals From Space Mean?
The search for intelligent life outside of Earth is a constant in science. Endless studies have been dedicated to searching for exoplanets with conditions for the development of life. But, in addition to biological markers, there is another way to search for traces of complex life in dark and deep space.
Researchers assume that, if humans have been able to develop wave-emitting technology, any intelligent civilization out there would have been able to develop advanced technology as well.
In recent years countless AI has been designed to search for radio signals coming from space. But even with the help of AI, there are many obstacles that the search for complex life beyond the planet faces.
The researchers explain in their article that, while the idea of AI searching for intelligent life is very attractive for science fiction stories, in real life, it is not due to the incorrect use of both concepts: neither the AI is intelligent, nor extraterrestrial intelligence searches can find direct evidence of intelligence.
The AI is responsible for searching for what it was designed for and is not capable of thinking for itself as to what it is detecting exactly. In the past, different studies that use radio telescopes to find signals of intelligent life in the Universe have faced a limitation: human technology itself.
Radio telescopes around the world, whose function is to capture signals from space, can not only detect radio waves produced by natural cosmic explosions but also radio signals and interferences produced by human technology are leaked. This is why, although signals are detected from the cosmos, it does not necessarily mean the existence of intelligent life.
AI to Search for Technological Signatures and Possible Extraterrestrial Life
With this problem in mind, the team of radio astronomers from the University of California Berkeley has developed a new AI designed exclusively to separate data produced by radio emissions from objects such as supernovas, and data produced by what they call technological signatures.
“Our AI was trained to search through radio telescope data for signals that could not be generated by natural astrophysical processes,” says Danny Price, researcher, and co-author of the study.
According to the researchers, the AI’s search algorithm is capable of quickly separating real technological signatures from ‘false positives.’ The deep learning system was devised by Peter Ma, who created a training dataset and then inserted simulated signals into real data to train the algorithm and turn it into an automatic encoder.
“As the automatic encoder processed the data, it ‘learned’ to identify standout features in the data,” says Daniel Price, co-author of the study. It is possible that for the AI to work optimally, it may have to go through an error threshold, and perhaps in the future, be refined to detect evidence of extraterrestrial technology.
Story originally published in Spanish in Ecoosfera.
