Could AI be Fooled in the Search for ET?

James Maynard
Jan 29 · 4 min read

Using artificial intelligence to search the skies for signs of alien life may present a unique set of problems not seen with human data analysis, a new study reveals.

As the Dawn spacecraft orbited the dwarf planet Ceres, the spacecraft took images of the Vinalia Faculae region on that diminutive world. When examined by artificial intelligence, the photographs appeared to show a dark triangle encompassing a bright square region. The problem was that this triangular feature is just an illusion created by several smaller markings seen on the surface of that world.

More than a minor illusion, this finding from the Spanish Foundation for Science and Technology (FECYT) suggests that instead of eliminating human error in the detection of patterns, artificial intelligence could have its own unique problems identifying patterns within chaotic data.

Using data from a spacecraft which explored the solar system, researchers set their sights on understanding the natures of biological and artificial intelligence, and finding what it could mean in the search for extraterrestrial life.

A New Dawn

A few bright patches are seen on the dark background of Ceres.
A few bright patches are seen on the dark background of Ceres.
Ceres, a dwarf planet orbiting in the main asteroid belt, is known for its sizable fields of salt and ice. Image credit: NASA

In 2007, NASA launched the Dawn spacecraft on a mission to explore the two largest members of the main asteroid belt. In 2011, Dawn arrived at Vesta, the second-largest member of the asteroid belt, becoming the first spacecraft ever to orbit any body between Mars and Jupiter.

After concluding its study of Vesta, Dawn took off toward Ceres, entering orbit around the dwarf planet four years later. In so doing, Dawn earned its title as the first (and so far, only) spacecraft to enter orbit around two bodies, not including Earth.

Bright white markings seen on the dark background of Ceres were shown to be composed of volcanic ice and salt.

Like the TIE fighters of Star Wars, Dawn was powered by an ion engine (although the fictional fighters first imagined by George Lucas had significantly more acceleration).

Dawn operated flawlessly until the robotic explorer ran out of fuel in 2018. During the course of its mission, the spacecraft carried out groundbreaking research adding to our collective knowledge of both dwarf planets (Ceres) and mammoth asteroids (Vesta).

A virtual flight over the dwarf planet Ceres, highlighting discoveries made by the Dawn spacecraft. Video credit: NASA/JPL

It’s like an Interplanetary Rorschach Test…

Researchers at the University of Cadiz in Spain wanted to know how artificial intelligence compared to human analysis of the features in the region. Researchers postulated that the same pattern recognition utilized examination of these geological features is similar to what is needed to identify technosignatures from alien civilizations.

“…[S]ome people seemed to discern a square shape in Vinalia Faculae, so we saw it as an opportunity to confront human intelligence with artificial intelligence in a cognitive task of visual perception… with implications bearing on the search for extraterrestrial life (SETI),” explains Gabriel G. De la Torre, a neuropsychologist at the University of Cadiz.

De la Torre assembled 163 volunteers with no significant knowledge of astronomy, asking participants to examine pictures of the region. Their observations were compared to those obtained by an electronic system, based on convolutional neural networks (CNN), previously “trained” to recognize triangles and rectangles.

“We wanted to investigate how the search for techno-signatures or oddities might be influenced by our cognitive skills and consciousness, and whether artificial intelligence could help or not in this task,” researchers wrote in an article detailing their study, published in Acta Astronautica.

Both human observers and the AI identified a square in the image, but electronic analysis also detected a triangle encompassing the square. When this feature was pointed out to the human participants, a significant fraction reported seeing the figure. In reality, the triangle was a haphazard series of small markings, which only have a passing resemblance to a single geological feature.

Researchers involved in this study caution that artificial intelligence employed in the search for extraterrestrial life may fall into a similar fallacy, finding patterns where none exist. Such an error could create false positives, signaling the presence of alien life where it does not exist.

“Look at you, hacker: a pathetic creature of meat and bone, panting and sweating as you run through my corridors. How can you challenge a perfect, immortal machine?”
― Ken Levine

Astronomers and astrophysicists are constantly poring through raw data looking for similar patterns that could reveal the presence of an unknown exoplanet, geological figures on other worlds, or the distinctive signal of alien life.

This study suggests that both humans and artificial intelligence will need to continue providing their own unique skills in the search for order in the midst of chaos.

James Maynard is the founder and publisher of The Cosmic Companion. He is a New England native turned desert rat in Tucson, where he lives with his lovely wife, Nicole, and Max the Cat.

Did you like this article? Subscribe to The Cosmic Companion Newsletter and never miss a story!

The Cosmic Companion

Exploring the wonders of the Cosmos, one mystery at a time

James Maynard

Written by

Writing about space since I was 10, still not Carl Sagan. Mailing List/Podcast:

The Cosmic Companion

Exploring the wonders of the Cosmos, one mystery at a time

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade