AI Debunks Claim That Ancient Tracks Belonged to a Predatory Dinosaur: They’re From a Herbivore

As has become customary, artificial intelligence has once again managed to overcome human limitations to solve a prehistoric mystery. 

A deep neural network AI identified the true origin of some ancient dinosaur tracks discovered in 1970. The footprints appeared to be from a fierce predator of the Cretaceous Period — at least to paleontologists — but they actually belong to a herbivorous ornithopod dinosaur. 

This prehistoric creature is estimated to have been nearly two meters long and as heavy as an Australovenator, a megaraptorid theropod dinosaur that lived between 100 and 95 million years ago. 

Researchers from the University of Queensland used this new technology to re-analyze the tracks that have been at the Dinosaur Stampede National Monument since the 1970s. Why? There are two fundamental reasons:

• These large tracks are surrounded by other tiny dinosaur footprints. Therefore, determining exactly which species made them tens of millions of years ago was quite difficult.

• Furthermore, even after 52 years, there were contradictory positions within the scientific community. Even within the research team, before AI arrived, one person believed the tracks were from a carnivore, while the other two were undecided about the dinosaur’s identity.

So, to truly verify who the “fierce predator” was, the team requested a deep learning artificial intelligence model.

What Did This AI Do With the Dinosaur Tracks?

This AI program, called “Deep Convolutional Neural Networks,” was trained on 1,500 dinosaur tracks of either theropod or ornithopod origin before analyzing the mysterious footprints. Therefore, all relevant dinosaurs from the Dinosaur Stampede National Monument, where the tracks were found, were registered.

After scanning the tracks and running through its gallery, the results were clear: they had been made by a herbivorous ornithopod dinosaur.

“We were quite stuck. When we used the AI on the large dinosaur tracks, all but one of these tracks were classified in the category of an ornithopod dinosaur, our prehistoric ‘predator.'”
Dr. Jens Lallensack

So AI once again surpassed the experts, this time unveiling the mystery of some ancient dinosaur tracks. 

Their research has already been published in the Journal of The Royal Society Interface, where we can see in detail this prehistoric discrimination between carnivores and herbivores. Perhaps the last we will witness, as the team plans to enrich the AI with new dinosaur tracks so as not to make such a mistake again in the future.

Reference:

A machine learning approach for the discrimination of theropod and ornithischian dinosaur tracks