There’s something almost magical about standing before a fossilized footprint left by a creature that roamed the Earth millions of years ago. But for paleontologists, that wonder often comes with a frustrating question: which dinosaur actually made this track? Was it a fearsome carnivore like Tyrannosaurus rex, a gentle plant-eating giant, or perhaps something even more unexpected? For over a century, these mysteries have stumped even the most seasoned experts. Enter DinoTracker, a newly released AI-powered app that is turning fossil identification from an art into a science—and it’s taking both researchers and dinosaur enthusiasts by storm.
Published in PNAS on February 1, 2026, and making waves across ScienceDaily and major science publications, DinoTracker represents one of the most exciting applications of artificial intelligence in paleontology to date. The concept is elegantly simple yet technically sophisticated. Grab your smartphone, snap a photo or sketch of a dinosaur footprint, upload it to the app, and within moments, DinoTracker delivers an analysis of which type of dinosaur likely left the impression. No specialized equipment, no PhD required—just curiosity and a camera.
What sets DinoTracker apart from traditional methods is its approach to learning. Rather than training on datasets where humans had already labeled which footprints belonged to which dinosaurs—a process that inevitably introduces bias and errors—the development team took a different path. Led by Dr. Gregor Hartmann at Helmholtz-Zentrum Berlin in collaboration with Professor Steve Brusatte at the University of Edinburgh, the researchers fed their AI nearly 2,000 real fossil footprint silhouettes without any pre-assigned labels. They then augmented this dataset with millions of simulated variations, accounting for the compression, erosion, and edge displacement that naturally occur over millions of years of geological transformation.
The AI discovered on its own that dinosaur footprints vary along eight key dimensions: how widely the toes spread, where the heel sits, the total contact area with the ground, how weight distributes across the foot, and several other subtle anatomical signatures. When tested against expert classifications, DinoTracker agreed with human paleontologists about 90% of the time, even for controversial specimens that have divided the scientific community for decades.
But the app isn’t just a validation tool for experts—it’s making genuine discoveries. One of the most tantalizing findings involves footprints over 200 million years old that the AI flagged as remarkably bird-like. If these tracks were indeed made by birds or their immediate ancestors, it would push back the origin of avian species by tens of millions of years, fundamentally reshaping our understanding of evolution. DinoTracker also shed new light on mysterious footprints from Scotland’s Isle of Skye, suggesting they may belong to some of the earliest relatives of duck-billed dinosaurs ever discovered.
Perhaps most exciting is that DinoTracker is completely free. The researchers explicitly wanted to democratize paleontology, giving anyone—from professional scientists working in remote field sites to children discovering curious impressions on family hikes—the ability to contribute to our understanding of prehistoric life. As Professor Brusatte puts it, this computer network might have identified the world’s oldest birds, which is “a fantastic and fruitful use for AI.”
In an era where AI often feels abstract or concerning, DinoTracker reminds us that these tools can also unlock ancient secrets buried in stone for hundreds of millions of years. Sometimes the most cutting-edge technology helps us see farthest into the past.
Leave a comment