Turn a radio telescope to the stars within the sky, and it’s immediately deafened. From pulsars to radio galaxies, and ionospheric disturbances within the atmosphere to radio-frequency interference (RFI) from our own technology, the sky is a cacophony of radio noise. And somewhere, amongst all that, may lie a needle in a haystack: a signal from one other world.
For over 60 years scientists have been scanning the skies within the seek for extraterrestrial life but have yet to seek out any aliens. Once you consider the sheer volume of search space — all those stars, all those radio frequencies — versus our limited searches up to now, then it’s little wonder we have not found ET yet. It’s a frightening task, especially for a human.
Thankfully, we have got some non-human intelligence to affix the search.
Related: AI is already helping astronomers make incredible discoveries. Here’s how
Using artificial intelligence (AI) is reaching critical mass, in our on a regular basis lives and in science, so it is not any surprise that it’s now being employed in Seek for Extraterrestrial Intelligence (SETI). We’re not talking about Skynet, or the machines from The Matrix movies, and even Star Trek: The Next Generation’s Data. The AI that’s so in vogue at present relies on machine-learning algorithms designed to do very specific jobs, even when it’s simply to consult with you on ChatGPT.
To clarify how AI is assisting in SETI, astronomer and SETI researcher Eamonn Kerins of the University of Manchester compares it to the needle in a haystack problem.
“You principally treat the information as if it is the hay,” Kerins told Space.com Space.com. “Then you definately’re asking the machine-learning algorithm to let you know if there’s anything in the information that won’t hay, and that hopefully is the needle within the haystack — unless there’s other stuff within the haystack too.”
That other stuff is generally RFI, however the machine-learning algorithm is trained to acknowledge all of the sorts of RFI we already learn about. Those signals — the familiar patterns of mobile phones, local radio transmitters, electronics and so forth — are the hay.
The training involves “injecting signals into the information after which the algorithm learns to search for signals which can be like that,” Steve Croft, an astronomer with the Breakthrough Listen SETI project on the University of California, Berkeley, told Space.com The algorithm learns to identify the patterns of those familiar signals and disrespect them. Should it spot something in the information that it hasn’t been trained on, then it flags this up as something interesting that requires a human to follow up on.
“There have been attempts recently at sifting through a number of the Breakthrough Listen data with a machine-learning algorithm,” said Kerins. “The information had already been combed through quite fastidiously previously by more conventional means, but yet the algorithm was still in a position to pick recent signals after being trained on the stuff that we learn about.”
This project was led by Croft and an undergraduate student, Peter Ma of the University of Toronto, who wrote the algorithm and put it to work analyzing data from 820 stars observed by the 100-meter radio telescope at Green Bank Observatory in West Virginia. The information, totaling 489 hours’ value of observations, contained thousands and thousands of radio signals, just about all of which were human-made interference. The algorithm checked each one in all them and located eight signals that didn’t match anything it had been trained on and which had been missed by earlier analyses of the information.
These eight signals seem to come back from five different star systems, although they is perhaps misleading. They have not been detected since — to see a signal repeat is probably the most basic requirement for a signal to be considered interesting in SETI — and they’ll probably transform more RFI. Nevertheless, even that is useful, because they could be used to coach the subsequent generation of machine-learning AI so similar RFI could be avoided in the longer term.
Machine learning algorithms could be divided into two camps. One is generally known as supervised learning, which is the teach-it-everything-you-know approach. Unsupervised learning is a little bit different, in that you simply just feed the algorithm the information and let it determine what is critical, with none human biases.
“With a totally unsupervised approach you only throw all the information in, stir the pot and let the algorithm figure it out by itself,” said Croft.
As a secular example, suppose you will have a dataset of images of tables and chairs, and you would like the algorithm to tell apart between them. In supervised learning, you train the algorithm on a lot of images which can be marked ‘table’ or ‘chair’. With unsupervised learning, the algorithm has to tell apart between the 2 by grouping things that look similar with none prior training — for instance, it would select anything with a back to be a chair, and anything with a protracted top to be a table.
Kerins highlights the instance of a project led by Adam Lesnikowski of NVIDIA, who’re famous for his or her graphics cards but which are actually leaders in artificial intelligence. Lesnikowski, joined by Valentin Bickel of ETH Zurich and Daniel Angerhausen of the University of Bern, used unsupervised machine learning in a test to see whether it could spot artificial objects on the moon. The algorithm was fed images from NASA’s Lunar Reconnaissance Orbiter, and it needed to determine what was a typical lunar feature, comparable to a crater or a rille, and what wasn’t. The test was successful — the algorithm picked out the Apollo 15 lunar lander on the surface of the moon.
The concept is that technological aliens could have already visited our solar system, and left probes or artifacts on the planets, moons or asteroids. It’s possible there may even be an lively probe watching us straight away.
“A few of my colleagues are very serious about the concept of getting orbiters with a machine-learning algorithm on board,” said Kerins. A spacecraft could survey planetary surfaces in our solar system to go looking for anomalies that may very well be alien probes, possibly thousands and thousands or billions of years old now. Because unsupervised learning has the advantage of with the ability to function in real-time, it might have the ability to evaluate each image before moving on without having to attend to send all the information back to Earth for humans to take a look at.
Actually, within the age of ‘Big Data’, machine-learning AI is the way in which forward and is now getting used extensively in astronomy and in SETI, with the potential to do things faster and higher than humans can.
“It’s definitely fast,” said Kerins. “The closest we will get with humans is thru citizen science projects.”
With machine learning algorithms, humans are still intimately involved. A signal might get flagged up by the AI as being intriguing, but it surely remains to be humans who must follow up and investigate. The algorithms aren’t that smart.
A time may come soon, nevertheless, after they are that smart. Researchers at places comparable to Google DeepMind have been pursuing artificial general intelligence, or AGI. Whereas the algorithms we have now today are very specific, AGI would have the ability to forged its hand to anything and learn and grow while it does. An AGI could rapidly speed up beyond the capability of human intelligence.
The chances for AGI transforming SETI are tantalizing. We have already seen how machine-learning algorithms designed to play games comparable to chess or Go! are developing strategies that befuddle the human experts whom the AI is thrashing in these games. An AGI could surely think of recent ways wherein to go looking for alien life beyond the confines of human biases and experience.
“It might have the ability to map out all kinds of possibilities for a way language and communication could be conveyed through signals,” said Kerins. “It would have the ability to eat vast astronomical catalogs and judge on optical strategies on how and where to look.”
Steve Croft echoes Kerins’ optimism. “I hope AI evolves to the stage where we will ask it to take the blinders off and picture, from every thing it knows about physics, biology, chemistry, exoplanets and technology, what it thinks ET is perhaps doing. It can probably give you some good ideas!”
That is if it might probably, and even will, tell us anything. The creation of an AGI will, in a way, by like creating an alien, one which could be very much unlike us and which we’d struggle to grasp.
“We’d find it very hard to directly communicate with it,” said Kerins. “We may need some hierarchy of translators, and at the highest of that hierarchy is an intelligence that will settle on much smarter ways to look in SETI. If it makes contact, then how does that filter right down to the biological intelligences, the silly guys, us?”
We’d get a version of Chinese whispers, where the relevant information is passed down through the hierarchy, getting simpler and simpler until we receive the dumbed-down version. The AGI may even withhold information that it deems could be too complicated for us to grasp. If AGI managed to make a SETI detection, we may not get the total picture.
That is speculation, though. Within the here and now, AI is a robust tool that’s accelerating our searches for ET. It is a sure thing that if we do discover a signal from one other world in the longer term, we’ll have AI to thank for it.