They train an AI to search for intelligent life and find something strange

Thanks to the latest advances in artificial intelligence, astronomers have more effective tools for the search for intelligent life. Or, at least, for the search for signs that may be an indication of it. This is precisely what a team of scientists from Curtin University has recently achieved, whose results have just been published in Nature Astronomy.

His artificial intelligence is capable of finding something called technosignatures. In other words, it analyzes the signals captured by radio telescopes, looking for those that do not correspond to any cosmic phenomenon. Then, it does a second screening, separating those that may be related to possible interference from the telescope itself.

Those that remain are those that, initially, cannot be explained in any way, so they could correspond to signals sent from technologies similar to ours. That is to say, that they could have their origin in another civilization of intelligent life, capable of investigating in space as we do. So far, the first tests of this artificial intelligence have found eight of these technosignatures. Does that mean there is intelligent life out there? They neither confirm nor deny, as it is something much more complex.

An algorithm trained to search for intelligent life

As with any other artificial intelligence algorithm, the one developed by these scientists it had to be trained before being put to use. For this, real data was taken and some simulated signals were introduced. These data were processed by the algorithm, while indicating whether it was some type of interference. This allows the machine to know what the main characteristics of a normal radio signal are. It is important to say that although it is called artificial intelligence, these algorithms do not think or are intelligent, they simply learn from the coincidences and differences in a huge sample of data.

Once the training was completed, data from 480 hours of observation from the Green Bank Observatory in West Virginia were entered into the algorithm. Once processed, the artificial intelligence returned 20,515 signals of interest. But not all of them would necessarily be related to intelligent life.

Photo by stephen leonardi

Artificial intelligence does not take astronomers’ work

Artificial intelligence saves astronomers work, but they are still needed in the search for technosignatures. In fact, the 20,515 signals detected by the algorithm had to be analyzed manually.

Thus, the authors of the investigation found that there were eight that could not be attributed to radio interference. Those would be the technosignatures that could be related to intelligent life.

Now, does that mean that this algorithm has managed to find intelligent life? Of course not. In fact, as explained in an article for The Conversation one of the authors of the study, Danny Price, two things can happen. On the one hand, that, in reality, these signals correspond to some type of rare interference that has not been cataloged and differs from what usually occurs normally.

This would indicate that it has not detected intelligent life. But, on the other hand, it could be that it did not hit any technosignature and that would not necessarily mean that there is no intelligent life. There could be civilizations capable of producing technology similar to ours, but they are too far away or for some reason have not yet encountered us.

So far, these technosignatures have not been detected again, so there is still no explanation for them. It will be necessary to continue studying them with new instruments, both those that already exist and those that are still being developed.

At least we now know that artificial intelligence provides the tools to search for these signatures faster and, as Price himself exemplifies, find that civilized needle in the vast haystack that makes up the universe.

View Hide summary