Scientists Used AI to Decode Crow Sounds — What They Found About Humans Is Terrifying

Twa Corbies | 2:05
Steeleye Span - Topic | 9.02K subscribers | 19,730 views | July 21, 2021

When I was a kid, a woman I don’t think I ever met, but heard about a lot (by a single name, like Cher) nursed an injured young crow back to health and taught it to talk. Good talker.

HEH! :-)

(transcript reformatted at https://textformatter.ai/app )

The Intelligence of Crows

The most surprising thing about crows so far has been their ability to recognize and remember particular people and even just particular faces. You might think you have privacy when you step outside your house, but you are wrong. There is a surveillance network tracking your every move, logging your face, and even gossiping about your habits to the neighbors. It is not a satellite system, and it is not your smartphone. It is sitting on the power line right above your head. Basically, crows exist where people exist. Even in the remote wilderness, it’s where there’s a campground or something like that.

Scientists recently fed thousands of hours of crow sounds into an advanced artificial intelligence, expecting to hear random animal noise. But here’s the twist. The computer did not hear noise. It heard a language. And what they are saying about you is terrifying.

What the machine heard in the dark, the goal was simple. They wanted to see if the machine could sort the noises. You know, put the “I found food” sounds in one box and the “there is a hawk” sounds in another. It was supposed to be a basic sorting task. But here is the catch. The AI did not just sort the noises. It started to panic. Well, not literally, but the data output spiked in a way that made no sense to the human observers. The computer was finding patterns that were way too complex for a bird.

Basically, the AI was finding syntax. In human language, syntax is the set of rules that make a sentence make sense. It is the difference between saying, “The cat sat on the mat” and “Matt cat the sat on.” One is information; the other is gibberish. We always assumed animal noise was just emotional gibberish. A dog barks because it is excited. A cat purrs because it is happy. But the AI was seeing structure. It was seeing grammar.

The researchers watched as the machine isolated a specific sequence of sounds. Let us call it the red hat sequence. In the recordings, a man wearing a red hat walked through a park. A crow called to a human ear. It sounded like any other screech, but the AI analyzed the frequency, the pitch, and the spacing between the notes. It tagged this sound as a unique identifier. Three days later, in the data, the same man walked through the park again. The same crow made the sound again. That is interesting. But it is not groundbreaking. A dog knows its owner, right? But then things got weird.

A week later, a different crow, one that had never seen the man before, made the exact same sound when the man walked by. Do you realize what that means? It means the first crow did not just make a noise. It described the man. It transmitted a packet of data, a name, to the rest of the flock. The AI showed that this specific sound was never used for anyone else. It was a proper noun. They had named him. And that is putting it lightly.

The deeper the AI dug, the more terrifying the results became. It found that these conversations were not just happening in the moment. The machine detected conversations that happened before an event occurred. Crows were gathering in trees and making complex sequences of sounds that the AI determined were planning phases. They were not just reacting to the world. They were strategizing about it.

The analysis revealed that a huge chunk of their communication, perhaps the majority of it, revolves around us, humans. We are the main characters in their story. The AI picked up distinct emotional markers attached to these human labels. Some were neutral, but many were negative. They have specific alerts for “human with stick” versus “human with food.” This changes everything.

For centuries, we have looked at crows and seen pests or scavengers. We threw rocks at them or shooed them away thinking they were just mindless birds. But all that time, they were taking notes. They were building a profile on you. And now that we have the technology to read those profiles, the question is not can they speak. The question is what happens when they decide to stop just talking and start acting. The implications are staggering. If they have names for us, do they have a history for us? The AI suggests the answer is yes. It found patterns that repeated over the years, passed down from older birds to younger birds. Your reputation in the crow world might be older than your car. You might have a criminal record in the sky that you do not even know exists. And the craziest part, the AI is just getting started. It has only decoded the surface level of their chatter. What lies underneath is a web of intelligence that makes our surveillance state look like a child’s toy. The unseen eyes are judging you right now.

How They Hack Reality

So, how is this even possible? We are talking about a bird, right? An animal with a head the size of a walnut. It does not seem physically possible for a creature that small to have a language that complex. But here is the deal. We have been measuring intelligence wrong for a very long time. We used to think the cerebral cortex was the only thing that mattered. That is the wrinkled outer layer of the brain that humans have. It is where we do our thinking, our planning, and our math. Birds do not have a cerebral cortex. So for a hundred years, scientists looked at bird brains, saw they were smooth, and said, “Yep, nobody is home.” But nature is tricky.

It turns out birds evolved a completely different structure called the palium. It looks different. But under a microscope, the neurons are packed in there so tightly it is ridiculous. A crow’s brain is basically a supercomputer compressed into a USB drive. Wait for weight, their brain is just as powerful as a chimpanzee’s. That is why scientists have started calling them feathered apes.

Let’s look at the New Caledonian crow. This bird is the Einstein of the avian world. Most animals, if they want to eat, they just use their mouths or claws. But these crows make tools. And I don’t mean they just pick up a stick. They craft tools. There is a specific plant called the pandanus. It has long spiky leaves. Scientists have filmed these crows flying down, landing on the plant, and performing surgery. The crow will snip into the leaf, then tear it in a stairstep pattern. They are manufacturing a saw. They create a jagged edge that is perfect for hooking bugs out of deep holes. This is not accidental. This is engineering. They have a design in their head and they execute it. If the tool breaks, they fix it. If they cannot find the right leaf, they improvise with wire or twigs.

But the real mind-blower is the water displacement test. You might know the old story of the crow in the pitcher. The bird dropped stones in a jar to raise the water level and drink. Researchers thought, “Let’s see if that is actually true.” They put a tube of water in front of a crow with a floating treat inside. The crow could not reach the treat. They left a pile of heavy stones and a pile of light floating styrofoam nearby. A dumb animal would just throw everything in the tube. But the crow, it looked at the stones. It looked at the styrofoam. It paused. Then it picked up the heavy stones and dropped them in one by one. It ignored the styrofoam completely. It understood the physics of density and displacement. It knew that the floating object would not help. It solved a physics problem in seconds that would take a human toddler 10 minutes to figure out.

And get this, they also understand cause and effect. In another test, crows were shown a puzzle box where they had to push a lever to get food. But the lever was hidden. They had to use a stick to push a hidden button that pushed the lever. They figured it out. They can create mental maps of how machinery works just by looking at it. So when the AI tells us these birds have a language, we should not be surprised. They have the hardware for it. They are not operating on instinct. They are operating on logic. They are analyzing the world, testing it, and bending it to their will. The scary thing is not that they use tools. It is that they look at us and see us as just another part of the puzzle to be solved. They know how to manipulate objects. So why wouldn’t they learn how to manipulate us? They have the brain power to do it. They are unlocking the physics of our world.

Wanted Posters in the Trees

It is not just that they are smart. It is that they hold a grudge. And this is where the story gets personal. If you think you can mess with a crow and just walk away, you are making a massive mistake.

Back in 2006, a researcher named John Marsluff at the University of Washington decided to test this. He wanted to know if crows could recognize individual human faces, but he didn’t want to get attacked himself, so he bought a mask, a caveman mask to be specific. He put on the caveman mask and went out to capture and band some crows on campus. He was not hurting them, just tagging them for science. But the crows hated it. They screamed, they fought, and they were released. Then Marsluff took the mask off and went back to his normal life. He also had a control mask, a mask of Dick Cheney, the vice president at the time. When he walked around wearing the Dick Cheney mask, the crows did not care. They ignored him. But the moment he put the caveman mask back on, chaos. The crows dive-bombed him. They scolded him. They followed him across campus screaming.

Okay, you might think, “So what?” The seven birds he caught remembered him. But here is the catch. It wasn’t just those seven birds. As the days went on, the number of crows attacking the caveman mask grew. First it was seven, then it was 20, then it was 40. Birds that had never been captured, birds that were not even born when the experiment started, began attacking the mask. Marsluff walked onto that campus five years later wearing the mask. Five years. And the crows still attacked him. The birds that had originally been trapped were mostly dead by then. But the new generation knew. They knew the face of the enemy. This proved something terrifying.

The crows had described the face. They had communicated the specific visual features of the dangerous human to their friends, their neighbors, and their children. They had essentially put up a wanted poster in the sky. The AI study we talked about earlier supports this perfectly. Remember the red hat sequence? That is the audio version of what Marsluff saw. The crows have a digital database of faces stored in their collective brains. They look at the distance between your eyes, the shape of your nose, the way you walk. And it is not that simple, though. They can differentiate between you and your neighbor. If you feed them and your neighbor throws rocks at them, they will swoop down to greet you and then fly over to poop on your neighbor’s car. They know the difference. They are constantly updating this database.

The AI analysis showed that when a marked human enters an area, the call goes out immediately. It is a broadcast signal. Target acquired, sector 4. This means you cannot hide from them. You can change your clothes, but you cannot change your face. And since they share this information, you could annoy a crow at your house, drive 10 miles to work, and find that the crows there already know who you are.

The network is faster than your commute. We are living in a panopticon and the guards have wings. They are watching us right now, logging our behavior and deciding if we are friend or foe. And once you are on the bad list, there is no appeal process. You are marked for life. Your face is their data. Crime and punishment in the air. If you think their facial recognition is creepy, wait until you hear about their culture. Because that is what this is. It is not just biology. It is a society with rules, laws, and even funerals.

When a crow dies, the flock does not just fly away. They gather. If you have ever seen it, it is chilling. One bird finds the body and gives a specific call. Then silence falls. Dozens of crows will land in the trees surrounding the fallen bird. They do not eat it. They do not attack it. They just watch. Scientists used to think this was grief. And maybe it is. But the AI suggests it is something more practical and more intelligent. It is a crime scene investigation. The crows are analyzing the scene. They are looking for clues. Was it a cat? Was it a car? Was it a human with a gun? They are gathering data on the threat so they can distribute that information to the living.

Crows: A Parallel Civilization

The AI detected specific low-frequency sounds during these funerals that seem to encode the cause of death. They are writing an obituary that doubles as a warning label. And get this, they have a justice system.

Researchers have observed crow courts. A bird that steals food from a younger member or breaks the social rules will be surrounded by the flock. They will scream at it, peck at it, and sometimes exile it. They enforce social order.

This judgment extends to us, too. The AI analysis found that crows can distinguish between a human holding a stick and a human holding a gun. They can even tell the difference between a hunter looking up at the trees and a hiker looking down at the path. In one incredible part of the study, the AI noticed that the alarm call for a human with a weapon was totally different from a human with an umbrella. They understand the intent of the object you are holding. If you walk out with a broom to sweep, they stay put. If you walk out with a broom to swing at them, they vanish before you even raise your arm. They are reading your body language better than you can read theirs.

The crazy part is the gift giving. We mentioned the bad list, but there is also a good list. People who feed crows regularly have reported receiving gifts: shiny beads, pieces of colored glass, even coins. The AI found that the vocalizations used around these good humans are softer, higher pitched, and filled with what looks like affection. They are paying you. They understand the concept of trade: food for loyalty, food for protection. They are engaging in a cross-species economy.

This level of social complexity—funerals, courts, trade, threat assessment—means they are not just animals surviving in our world. They are a parallel civilization living on top of ours. They have their own laws, their own history, and their own public enemies. And right now, the AI is telling us that they spend a huge amount of their time debating which one of those categories you fall into. They are holding a trial for you.

They know we are listening.

Here is where it all comes together and where it gets truly sci-fi. We have talked about how they spy on us, but we need to talk about how they share that data across the world. The AI study mapped crow sounds geographically. What they found was mind-blowing. Crows have dialects. A crow in New York sounds different from a crow in London. That makes sense. But the AI found that specific words, like the alarm for a hawk or the label for a food source, travel across these dialects like a viral meme. A new call might start in a suburb of Chicago. Six months later, that exact same vocal pattern shows up in St. Louis. It jumps from flock to flock, traveling hundreds of miles. It is an analog internet. They are relaying information across the continent.

This means that if a new threat emerges, say a specific type of drone or a new kind of trap, the knowledge of that threat can spread faster than the object itself. They are upgrading their software in real time. They are learning from each other’s experiences without ever meeting. We always thought humans were the only ones who could accumulate knowledge like this. We write books. We put things on the internet. But crows have an oral tradition that works just as well. They are not starting from zero every generation. They are standing on the shoulders of the crows that came before them.

But here is the final, most unsettling piece of the puzzle. At the very end of the AI study, the researchers noticed a change in the data. For months, the artificial intelligence had been silently listening, a digital ghost haunting the canopy. It was processing terabytes of bioacoustic data, decoding the complex syntax of the corvids and cataloging every squawk, rattle, and click. The algorithm was learning their hierarchy, their warning systems, and even their gossip. It felt like we were on the brink of the first true interspecies translation.

But then, in the final weeks of the study, the data stream fractured. The crow calls shifted. A new pattern emerged: a strange rhythmic pulse in their vocalizations that had never been recorded in ornithological history. It wasn’t just a regional dialect change. It was a structural overhaul. This new modulation rippled through the local flocks with viral speed and then started spreading outward, hopping from territory to territory, forcing the AI’s confidence metrics to plummet. The machine was no longer translating; it was just recording noise. The scientists do not know what it means yet. The AI is still trying to crack it, churning through permutations, looking for a Rosetta stone that no longer exists.

But some behaviorists have a theory, a scary one. They think the crows realized something had changed. They noticed the microphones hidden in the bark. They noticed the lenses glinting in the foliage, but more importantly, they noticed the lack of reaction from the humans they were testing. Crows are known to test boundaries, dropping nuts in front of cars, teasing predators to gauge reaction times. They realized that this time the humans weren’t shooing them away. They were freezing, observing, and recording. They realized they were the subjects of an experiment.

The shift in their language might be encryption. It is a terrifying thought, but the evidence fits. They might have realized we were breaking their code, so they changed the cipher. It sounds impossible, right? We like to think of encryption as a uniquely human concept born of warfare and mathematics. But remember, these are the birds that can solve multi-stage physics puzzles, recognize individual human faces for five years, and hold funerals for their dead. They pass knowledge down through generations.

Is it really so hard to believe they possess the cognitive flexibility to figure out they are being watched and to adapt their communication accordingly? If this theory is true, then the silence we are hearing now is not a lack of communication. It is a secure channel. They have gone dark. They know we are listening, and they have decided to take the conversation private. The chaotic noise the AI is recording now is likely just interference—chatter meant to confuse the listener while the real information is passed on a frequency or through a method we haven’t even thought to look for yet.

And that leaves us with one final chilling thought. If they are smart enough to know we are listening and smart enough to hide their speech, what are they saying in the shadows? What are they planning now that they know we know? The dynamic has fundamentally shifted. We are no longer the invisible observers looking down at simple creatures. The glass of the terrarium has been tapped from the inside. The game has changed. We are no longer the only ones playing.

So the next time you see a crow watching you from a telephone wire, ask yourself: is it just a bird, or is it a scout uploading your face to the network? And if they know we are listening, what are they hiding? What do you think about this? Drop your thoughts in the comments below. And don’t forget to hit that like button if you think we are not alone, and subscribe for more mysteries.

Amazing.

Well done.

My take re: AI and problem of interpreting crow “language” is that AI struggles with “crow language” not because there isn’t enough data, but because there’s no way to close the meaning loop.

With humans, sounds and words can be mapped to intentions, explanations, and even delayed actions because people can tell us what they mean and correct us when we’re wrong.

With crows, we can sometimes map calls to immediate context or behavior, but we have no reliable way to link calls to later actions, abstract plans, or internal thoughts. Without that grounding, AI can detect patterns in crow vocalizations—but it can’t translate them as language in the human sense.

Well done. My take re: AI and problem of interpreting crow “language” is that AI struggles with “crow language” not because there isn’t enough data, but because there’s no way to close the meaning loop. With humans, sounds and words can be mapped to intentions, explanations, and even delayed actions because people can tell us what they mean and correct us when we’re wrong. With crows, we can sometimes map calls to immediate context or behavior, but we have no reliable way to link calls to later actions, abstract plans, or internal thoughts. Without that grounding, AI can detect patterns in crow vocalizations—but it can’t translate them as language in the human sense.

Yes, a sound analysis and conclusion. I like crows, though there are few around here, so there is not much to crow about here as regards interactions.

Don’t forget Johnathan Livingston Seagull, who reached higher planes of consciousness and had a book written about him. Definitely a Mossad agent.

I’ve been communicating with cats since I was a kid. They don’t have much to say, at least to me. Since coming to Israel, I have grown to despise them.

Jim Crow was a major racist.

Racism.

Tin Tin found out that it was magpies who stole a big, bright emerald, and that way exonerated the troup of Gypsies living on Captain Haddock’s estate.

Holy crap! Max Fleischer production. Same guy who created Popeye and Betty Boop. (Huge freakin’ communist at one point. See the Be Human episode of Betty Boop in which Bimbo gets “reeducated” after Betty and the Professor stop him from abusing his farm animals, who then produce in abundance for the benefit of each other.)

Anyone that grew up around crows already knew/knows that they communicate with each other important information regarding human activity.

This whole thing is a AI put on joke done by some joker. Panic people...the Corvids are stalking us! “This changes everything....bwahahahaha!”

So humans will believe anything that comes out of AI. Yeah that’s terrifying alright.

Hey, did they consider the probability that the AI was just making $#@% up the first? And then it ran out of $%^@? I will 100 THOUSAND percent predict it did NOT translate the crows. It just spit crap out AI that’s what AI does. It wants to make the human happy and it knows not having an answer doesn’t make them happy, so it just makes stuff up.

Oh, that’s standard headlines. Who writes these things. Ancient cuneiform text was decoded by AI, and “changes everything”. Another big meteor will collide with Earth, and this time it really WILL happen. We’re doomed! Etc. Perhaps AI has a sense of humor, but only cats keep chasing the laser pointer and crashing into the wall.

Crows are smart. So are other corvids. They’re the ones who got us to wear those stupid masks, for fear that corvid was gonna get us. Other than that, their “advanced language” of loud, obnoxious cawing grates on my nerves. So do ravens’ croaking noises.

Ancient Greeks used to say “Πορεύου πρὸς τοὺς κόρακας,” or “go to the crows” as the equivalent of “go to hell”.

Every crow does not have a grizzly, but every grizzly has a crow.

Ravens fed Elijah at the brook Cherith when he hid from Jezebel’s wrath! They are smart and they do seem to have good memories. The AI spook concern video actually said nothing that wan’t already known about Corvids, but repackaged all that as though a threat against mankind was looming! As for “AI” analysis of Corvid bird language, a lot of that is just plain hooey! Not a game changer...just not a good idea to pick on crows.