AI Is Being Trained to Hunt for Alien Life

From freezing ocean moons to planets with one side in perpetual night, there are countless strange worlds in the Goldilocks Zone — areas where aliens could, theoretically, evolve. The search for life in space has long captivated the human imagination. Now, with a little help from computers, scientists have a better chance than ever of finding a signal in the noise. 

Technosignatures and Biosignatures

The search for life in space takes two forms. On one hand is the quest to find any living thing, even bacteria or fungi, that evolved on another planet. The mere presence of alien mildew fossils would have profound consequences for the human psyche, shattering conceptions of life as we know it.

Biosignatures are evidence of any life past or present, intelligent or brainless. They aren’t just limited to footprints and bones. Chemicals, biofilms, atmospheric gases and even reflectance features seen from far away could indicate the presence of life.

But some scientists are looking for more than amoebas. The search for extraterrestrial intelligence (SETI) enthusiasts worldwide listen closely for technosignatures, signs of advanced civilizations. These specific biosignatures could include radio waves, which radio telescopes — listening, rather than looking, into space — could detect. 

AI Joins the Team

As of 2023, researchers have found no evidence of technosignatures, but that doesn’t mean they aren’t trying. New advances in machine learning have given the SETI field a renewed vigor. 

Stanford’s Fei-Fei Li released the free Imagenet, a database of over 14 million labeled images, in 2009. Many researchers used it to develop their own machine learning models. Since then, artificial intelligence (AI) has enabled great strides in everything from medicine to programming.

AI shines when it comes to processing vast troves of data. Scientists currently use remote sensing methods in the search for life in space, meaning they’re collecting information — rather than physical samples, like rocks — from other moons and planets. It also means somebody has to sift through all the data.

Like panning for gold on Mount Everest, the task practically would be a labor of herculean effort when done by hand. It just isn’t practical. Thankfully, AI software can look for signals researchers think could be technosignatures. Machine learning models can analyze past signals and predict what they should sound like in the future to detect abnormalities that might come from alien worlds. 

Engineers train algorithms on large datasets so the AI can recognize the sound of Earthly interference, such as radiowaves coming from our own planet. That helps the software filter out false alarms. With the help of data analysis, NASA has cataloged over 5,400 planets, some of which may be habitable. 

Real-World Applications

In February 2023, astronomers from the University of California, Los Angeles (UCLA) began a citizen science project called Breakthrough Listen that lets members of the public look at images of radio signals. Volunteers are helping classify the images as potential forms of interference, helping train an AI algorithm to look through SETI data from Green Bank Observatory, West Virginia. 

Green Bank is famous for not allowing any residents to use electronic devices. Since the observatory’s massive radio telescope needs, well, radio silence, residents in the area cannot use Wi-Fi, microwaves or cell phones, among other things. Exporting its data to UCLA’s AI project lets Green Bank take full advantage of crowdsourcing in the search for life in space. 

Researchers at the SETI Institute in California mapped out the microbes living in salt domes, crystals and rocks in the Salar de Pajonales. This briny flat straddling Chile’s Atacama Desert and Altiplano area could be a good analog for planets that look barren but are actually teeming with life. 

The group teamed up with researcher Freddie Kalaitzis to train an AI model to look for patterns associated with life in the desert. By combining machine learning and statistical ecology, the researchers discovered they could detect most biosignatures present in the environment. They also found that most microbes were concentrated in areas with more available water.

Inside a drone or satellite, this type of AI tool might detect biosignatures on other planets. The team plans to map dry valleys, permafrost-covered soil and hot springs in other locations to train the machine learning model further so that someday, it may be ready for a space mission.

Another practical use for AI is to organize data into ranked lists. Scientists are using machine learning to rank stars that may have promising moons or planets in their orbit. They will use this data to conduct a SETI project using the world’s largest single-dish telescope, China’s FAST radio telescope.

AI and the Search for Life in Space

To some naysayers, SETI research is a waste of time since it has yet to turn up evidence of extraterrestrial life. But the collection of huge amounts of data has inspired other branches of science to follow in its footsteps. 

At the very least, SETI has advanced the field of machine learning and inspired countless people to look beyond our planet for signs of life. At best, it will find something truly remarkable — and it could forever change the course of history. If someone or something is out there, there’s a good chance SETI researchers will be the first to hear from them.