AI Helps Bring Back the Dodo

Colossal Biosciences is working to revive the dodo from extinction.

There are many reasons why there are so many concerns about the harm that AI can cause, which explains why the technology has so many opponents. But even the harshest critics may be smiling at the news from Dallas-based Colossal Biosciences. There, researchers have just announced that artificial intelligence may help speed the return of the extinct dodo, with the flightless bird that disappeared from the world more than 350 years ago and is likely to walk among us again in the early 2030s. With so few ancient samples of dodo DNA degraded over the centuries, the numerical and analytical power of artificial intelligence is what is needed to fill in the genetic gaps.

"Extinction is as much a data problem as it is a biological problem," Colossal CEO and co-founder Ben Lam said in a statement. [AI] learns how regulatory DNA shapes living birds, then identifies genetic switches that may contribute to the dodo's distinctive anatomy.

In 2025, when Colossal announced that it had sequenced the ancient DNA of the extinct direwolf, it used it to rewrite the genome of an existing gray wolf, bringing the Ice Age direwolf back to life more than 10,000 years after its disappearance. Animals also on the company's list for extinction include the woolly mammoth, the Tasmanian tiger, the ostrich-like moa and the antelope-like quagga. The dodo, with its waddling gait, stilettos, googly eyes and ridiculous beak, has none of the power, grace or charisma of those other animals - but that's part of its charm.

In his 1941 book How to Be Extinct, New York Herald Tribune writer Will Cappy wrote, “The dodo never stood a chance. "He seems to have been invented for the sole purpose of becoming extinct, and that was all he was good for. I don't blame the dodo, but he was just a mess. The Portuguese called the birds dodos, or fools, because they were friendly with the Portuguese and allowed themselves to be clubbed."

But what inspires scorn also inspires love—and it's going to take a lot of it to bring Dodo back. Eliminating the extinction of a paired mammal like a dire wolf is a relatively simple process. The massive researchers first harvested cells from a living gray wolf and made 20 edits to 14 genes that code for the wolf's white coat, larger size, stronger jaws, cold-weather metabolism, and more. They then extracted the nucleus of that transcribed cell and inserted it into a denucleated gray wolf egg. The fertilized egg was then allowed to develop into an embryo and implanted in the uterus of a domesticated greyhound. A little over 60 days later, that mother dog gave birth to a wolf cub.

It will be more difficult to eliminate the extinction of a bird that carried out its pregnancy outside the womb. For the dodo, the researchers will first harvest primordial germ cells (PGCs)—the nascent cells that produce eggs and sperm—from the Nicobar pigeon, the dodo's closest living relative. PGCs are then rewritten to add dodo features, in the same way that gray wolf was edited. These cells are then injected into microscopic chick embryos through the shell of domestic eggs. PGCs migrate to the embryonic gonads of the future chick and reside there. When the edited female chick hatches and grows and mates with an edited male, she lays eggs that produce not a chick like her, but a dodo.

"In an ideal world, if we had hundreds of Nicobar pigeons [eggs], we would use pigeons as surrogates," says Anna Keith, Colossal's bird species manager. "But Nicobars are seasonal breeders. They only lay a few months a year and usually only lay two eggs in a clutch. Eggs are very easy to get. Some chicks lay one egg a day. You can only get so many more replacement eggs from Nicobars than one chick."

But editing PGCs in the first place would be no small task, simply because there are so many places in the dodo's genome that encode characteristics of the species, especially its ice cream-shaped beak. It's not exactly the genes the researchers read in their work, but the regulatory elements, or parts of the genome, that determine which genes are turned on and off. And for the beak alone, that's a heavy load. There are over 650 regulatory elements associated with beak size and nearly 1,700 with beak thickness and width. Worse, DNA fragmentation in ancient samples provides incomplete records of which of those regulatory switches should be flipped. A new artificial intelligence platform that the company calls GenoMorph can fill these gaps.

To develop GenoMorph, Colossal scientists created an atlas that shows how the genetic switches that control facial development are turned on and off during embryonic development in chickens, quails, ducks and pigeons. They then compared the genomes and skull shapes of more than 70 bird species, including the dodo's closest living relative, the Nicobar pigeon. The result is a list of about 2,350 genetic switches that might help explain the dodo's unusual beak, which the team is now testing one by one. This work builds on a separate achievement: a remarkably complete dodo genome, assembled from a well-preserved specimen using the Colossal software originally developed for its dire wolf project.

"For an ancient genome, this gene is extraordinary," says Beth Shapiro, Colossal's chief scientific officer. "There are a few small holes, but much less than what we usually see in ancient DNA. We're not really missing much of the dodo genome."

The beak may be the dodo's most dramatic feature, but there are plenty of other features spelled out in the two billion letters that make up the dodo's genome — which is why giant scientists wait five or more years to bring a dodo back into the world.

In the same way that some of the work Colossal did to read and analyze the dire wolf's genome helped later develop GenoMorph to read the dodo's genome, the dodo's work could similarly lend itself to reading and editing the genomes of other species to become extinct.

"It's exciting to see that the approach that was developed for the dire wolf is now creating value in another species program," Shapiro said in a statement. "Each scientific breakthrough becomes part of a scalable technology platform that can accelerate subsequent applications." For now, Dodo remains a biological vaporware—but it's a vaporware that could take shape before long in the 21st century.