There are a whole lot of reasons to have a whole lot of concerns about the harm artificial intelligence can cause, which explains why the technology has so many detractors.
asserted
technology → be → detractors
But even the harshest critics might smile at the news coming out of Dallas-based Colossal Biosciences.
uncertain
critics → smile → Biosciences
There, researchers have just announced that AI may help hasten the return of the extinct dodo, with the flightless bird, which vanished from the world more than 350 years ago, likely walking among us again sometime in the early 2030s.
uncertain
which → announce → 2030s
With so few ancient samples of dodo DNA surviving—and that which does exist having degraded over the centuries—the analytic and number-crunching power of AI is just what’s needed to fill in the genetic blanks.
asserted
what → survive → blanks
“De-extinction is as much a data problem as a biology problem,” said Colossal CEO and co-founder Ben Lamm in a statement.
asserted
Lamm → say → statement
“[AI] learns how regulatory DNA shapes living birds, then identifies the genetic switches that may have contributed to the dodo’s distinctive anatomy.”
uncertain
that → learn → anatomy
Colossal established its de-extinction bona fides in 2025, when it announced that it had sequenced ancient DNA from the long-extinct dire wolf, used that to rewrite the genome of the extant gray wolf, and brought the Ice Age dire wolf back to life, more than 10,000 years after it vanished.
asserted
it → establish → life
Animals that are also on the company’s de-extinction to-do list include the woolly mammoth, the Tasmanian tiger, the ostrich-like moa, and the antelope-like quagga.
asserted
that → include → mammoth
The dodo, with its waddling gait, its stubby wings, its goggly eyes, and its preposterous beak, has none of the power, grace, or charisma of those other animals—but that’s part of its charm.
asserted
that → have → charm
“The dodo never had a chance,” wrote New York Herald Tribune writer Will Cuppy, in his 1941 book How to Become Extinct.
asserted
Cuppy → have → book
“He seems to have been invented for the sole purpose of becoming extinct and that was all he was good for.
asserted
he → seem → purpose
I’m not blaming the dodo, but he was just a mess.
asserted
he → blame → dodo
The Portuguese called the birds doudos, or fools, because they would associate with the Portuguese on friendly terms and allow themselves to be hit on the head with clubs.
asserted
themselves → call → clubs
But what inspires mockery inspires love too—and it will take plenty of that to bring the dodo back.
asserted
it → inspire → dodo
De-extincting a placental mammal like a dire wolf is a comparatively straightforward process.
asserted
extincting → De → wolf
Colossal researchers first harvested cells from a living gray wolf and made 20 edits to 14 genes, coding for the dire wolf’s white coat, larger size, more powerful jaws, cold weather metabolism, and more.
asserted
researchers → harvest → coat
They then extracted the nucleus of that rewritten cell and inserted it into a gray wolf ovum whose own nucleus had been removed.
asserted
nucleus → extract → ovum
Next, the fertilized egg was allowed to develop into an embryo and was implanted in the womb of a large domestic hound.
asserted
egg → fertilize → hound
Just over 60 days later, that mama dog gave birth to a puppy wolf.
asserted
dog → give → wolf
De-extincting a bird, which does its gestating outside of any womb, will be harder.
asserted
which → De → womb
For the dodo, the researchers will first harvest primordial germ cells (PGCs)—nascent cells that give rise to eggs and sperm—from the nicobar pigeon, the dodo’s closest living relative.
asserted
that → harvest → pigeon
The PGCs will then be rewritten to add dodo traits, in the same way the gray wolf was edited.
asserted
wolf → rewrite → way
Those cells will then be injected through the shell of a domestic chicken egg into a microscopic chick embryo.
asserted
cells → inject → embryo
The PGC’s will migrate to the embryonic gonads of the chick-to-be, taking up residence there.
asserted
PGC → migrate → residence
When the edited female chick hatches and grows and mates with an edited male, she will lay eggs that will produce not a chicken like her, but a dodo.
asserted
that → edit → her
“In an ideal world, if we had hundreds of nicobar pigeon [eggs] we would use the pigeon as the surrogate,” says Anna Keyte, Colossal’s avian species director.
asserted
Keyte → have → surrogate
They only lay for a couple of months out of the year and they usually lay just two eggs in a clutch.
asserted
they → lay → clutch
Chicken eggs are super easy to get.
asserted
eggs → get → ?
Some chickens will lay an egg a day.
asserted
chickens → lay → egg
You can just get vastly more surrogate eggs from a chicken than you can from a nicobar.
asserted
you → get → nicobar
But editing the PGCs in the first place will be no small thing, simply because there are so many sites on the dodo genome that code for the species’s traits—especially its ice-cream-scoop-shaped beak.
asserted
that → edit → traits
It’s not precisely the genes that the researchers are reading in their work, but the regulatory elements—or the parts of the genome that determine which genes get switched on and off.
asserted
genes → ’ → genome
And for the beak alone, that’s a heavy lift.
asserted
that → ’ → beak
Worse, the fragmentation of DNA in the ancient samples gives an imperfect record of which of those regulatory switches are to be flipped which way.
asserted
which → give → switches
A new AI platform the company calls GenoMorph can fill in those gaps.
asserted
company → call → gaps
To develop GenoMorph, Colossal scientists built an atlas showing how the genetic switches that control facial development turn on and off as embryos grow in chickens, quail, ducks, and pigeons.
asserted
embryos → develop → chickens
They then compared genomes and skull shapes from more than 70 bird species, including the dodo’s closest living relative, the Nicobar pigeon.
asserted
They → compare → relative
The result is the list of about 2,350 genetic switches that may help explain the dodo’s unusual beak, which the team is now testing one by one.
uncertain
team → help → one
That work builds on a separate achievement: a remarkably complete dodo genome, assembled from a single well-preserved specimen using software Colossal originally developed for its dire wolf project.
asserted
Colossal → build → project
“For an ancient genome, this one is extraordinary,” says Beth Shapiro, Colossal’s chief science officer.
asserted
Shapiro → say → genome
…and 7 more, not listed.
The Aporia
