Library/Spotlight

Back to Library
TED TalksCivilisational risk and strategySpotlightReleased: 2 Jun 2025

Can AI help us speak with wolves? | Jeffrey T. Reed

Why this matters

Auto-discovered candidate. Editorial positioning to be finalized.

Summary

Auto-discovered from TED Talks. Editorial summary pending review.

Perspective map

MixedGovernanceMedium confidenceTranscript-informed

The amber marker shows the most Risk-forward score. The white marker shows the most Opportunity-forward score. The black marker shows the median perspective for this library item. Tap the band, a marker, or the track to open the transcript there.

An explanation of the Perspective Map framework can be found here.

Episode arc by segment

Early → late · height = spectrum position · colour = band

Risk-forwardMixedOpportunity-forward

Each bar is tinted by where its score sits on the same strip as above (amber → cyan midpoint → white). Same lexicon as the headline. Bars are evenly spaced in transcript order (not clock time).

StartEnd

Across 8 full-transcript segments: median 0 · mean -1 · spread -80 (p10–p90 00) · 0% risk-forward, 100% mixed, 0% opportunity-forward slices.

Slice bands
8 slices · p10–p90 00

Mixed leaning, primarily in the Governance lens. Evidence mode: interview. Confidence: medium.

  • - Emphasizes governance
  • - Emphasizes safety
  • - Full transcript scored in 8 sequential slices (median slice 0).

Editor note

Auto-ingested from daily feed check. Review for editorial curation under intake methodology.

ai-safetyted-talks

Play on sAIfe Hands

On-site playback is enabled when an episode-level media URL is connected. This entry currently points to a source page.

This entry currently has a show-level source URL, not an episode-level media URL.

Episode transcript

YouTube captions (TED associates this talk with a public YouTube mirror) · video saTGDtFQAMI · stored Apr 10, 2026 · 215 caption segments

Captions are an imperfect primary: they can mis-hear names and technical terms. Use them alongside the audio and publisher materials when verifying claims.

No editorial assessment file yet. Add content/resources/transcript-assessments/can-ai-help-us-speak-with-wolves-jeffrey-t-reed.json when you have a listen-based summary.

Show full transcript
So good morning, everybody. My name is (howls). But I'm about to play you a real wolf's howl. Not just any old wolf. This wolf was called 907. The typical wolf in Yellowstone National Park only lives three and a half years. 907 was 11 and had one eye when she howled like this for over 30 minutes to her pack mates five miles away. Close your eyes and just listen. (Wolf howls) (Wolf howls) (Wolf howls) OK, you can open your eyes. Here's what she said. With the help of AI and images like this, we're trying to decode the wolf howl and rethink how to protect and promote the world's last wild places. This was 907. She was the matriarch of the Junction Butte pack. But I like to call her the “One-Eyed Wonder.” And this was me 50 years ago with my family in Yellowstone. Look closely. There are three reasons you pee your pants. One -- (Laughter) You're getting old. Like none of us in this room, right? Two, you're too young to know better. Or three, in my case, it was my first experience watching wildlife in one of America’s grandest ideas: Yellowstone National Park and public lands. If you've never been to Yellowstone, it's that pale yellow dot on our pale blue planet where my buddies and I grew up hiking, hunting and fishing the Absaroka Mountains of Montana youth. And where whiskey is for drinking, but water and wolves are sometimes for fighting. Wolves were killed off in Yellowstone in the 1920s, reintroduced in 1995, and now contribute to over a half- a-billion-dollar annual tourism economy where I live. Now recently, the media has been talking a lot about the use of artificial intelligence to decode animal communication. From elephants to bats and everything in between. And a few of us are doing it with one of the most iconic sounds of the wild. (Wolf howls) In order to decode wolfish, my company works with Yellowstone scientists to build autonomous recording units, or ARUs. These battery-operated devices use AI to record only the sound and motion that matters to the scientists, saving them time and money. But they can do a lot of other things, like tell me the precise location of a sound, for example, the where and when of a poacher's gunshot. Last year, researchers hiked into the backcountry to place ARUs all across the Greater Yellowstone, including one near this wolf den site to study wolf communication. Wolves have over 20 different call types, including barks, yelps, whimpers, whines, moans, woas and even teeth clacking. And they're one of a few species who collectively communicate in what’s called a “chorus howl.” Now the pups first hear this dictionary of sounds at their birth dens. And so we departed. And one month later, the pack arrived and the pregnant alpha female claimed the den. Can you see her? Look real close, it's just a portion of her. There she is. Her nose is poking out of the den, and she's taking a well-deserved pregnancy nap. Now three months later, I created this visual representation of every sound recorded at the den site for the month of July. Each row represents one day of sound from midnight on the left edge to midnight on the right. And AI looks for patterns in this data, and the ARUs don't sleep at night, so they were able to record all of these wolf howls. Now those blobs of color you’re seeing are the birds starting to sing their dawn chorus each morning. And it shifts to the right throughout the month. Why? Because daylight is decreasing, and we're learning that wolves mostly howl at night. But a howl isn't just a howl. These are spectrograms of 50 individual wolf howls. A spectrogram simply visualizes sound, and your eyes and AI can see patterns in them. Here's the howl from just one of those wolves. (Wolf howls) And here's all 50 wolf howls at the same time. (Wolves howl) Software algorithms help us find signal in that noise, and researchers have already determined that wolves can identify one another just from their howls, but we do not know yet if they have names like Teddy or Rachel, or if it's more like you picking up the phone and recognizing that person's voice on the other end. Now a howl basically means "Here I am. Where are you?" But we've found five functions for different types of howls, including this one, which is used in contexts of distress. (Wolf howls) Now this is where it gets interesting to me. A chorus howl is when a pack or family of wolves communicate as a group, often to signal that this is their territory. It's kind of like a political rally. And in this video, a wolf from the pack is off camera howling, but being totally ignored by his pack mates. Until something happens. (Wolf howls) As soon as the alpha female howled, the rest of the wolves looked at her, and the chorus began. (Wolves howl) She was the conductor of this chorus. And as it kicks off, the wolves come together in this mosh pit of dancing bodies, making different types of calls. But some of the wolves, especially females, use a particular type of howl called a woa. And we think it's to strengthen family bonds. It's a wolf's version of "I love you," but it sounds like this. (Wolves woa) The entire event lasts about 60 seconds, and then everybody goes about their business. Now this is what a chorus howl looks like when 11 wolves from the same pack howl at the same time. And I don't know about you, but to me it looks like an angry doctor's signature. But not to the wolves. We've learned that wolves from one pack can count or estimate, the number of wolves howling in a rival pack without seeing them, and they make a decision. If they're outnumbered, they typically run for the hills. If not, they often approach that rival pack. We're using technology from Google DeepMind to see if we can count wolves just from audio recordings and improve census estimates of wildlife. Now wolves also bark, like your dog. But unlike most dogs, wolves combine barks with other sounds into "sentence-like" structures. And in this video, a female wolf uses a bark howl after being chased and nearly killed by a rival pack. The bark's an alarm, and loosely translated, means “stranger danger,” and the modulated howl, “send backup.” (Wolf bark-howls) (Wolf bark-howls) These secretive devices are teaching us that wolves communicate a lot in close quarters. Now we recently put a tiny ARU on the first wild wolf ever and learned that she likes to vocalize roughly half as much as a typical American speaks in a given day. And here's just a sampling of the sounds they make. For those of you with a dog, nod or shake your head if you recognize one. (Wolf yelps) (Wolves vocalizing) I love those mini howls. They make a lot more of those than we ever knew. They're close-range contact calls. So remember the alpha female taking a nap by her den? Twenty-five days later, her pup stepped outside the den, raised its tiny three-inch snout, and howled beside mom. (Wolves howl) With the help of AI, we're learning more about why wolves howl, including that your dog might be barking at that emergency vehicle because it matches the pitch of a wolf pup. Now look, I don’t know if AI will ultimately decode animal communication. Wolfish is not the same thing as English. But I do know this, there has to be animals to decode. If your body represented the total weight of all the world's land mammals today, your right forearm would be what's left of the wild ones. The rest of your body is us, our livestock and our pets. As for carnivores, things like lions, tigers and wolves, it's less than my pinky. The challenge we collectively face as real humans, not artificial ones, goes far beyond individual opinions on wild wolves. It's about the future of wildness itself. For hunter like me and non-hunter alike. So we're going to close with a group chorus howl. No practice. And the next time you're at a party, I want you to down a shot of whiskey and try this. Now here’s a tip. Use the vowel "oo" as in "boot." Wait till I count to three, and I'm going to tune up your voices with this recently rediscovered recording from 1995. These are the first wild wolf howls in Yellowstone after 70 years of silence. (Wolves howl) Ready? One, two, three. (Audience howls) (Howling ends) (Laughter) Well done. Thank you. (Applause)

Counterbalance on this topic

Ranked with the mirror rule in the methodology: picks sit closer to the opposite side of your score on the same axis (lens alignment preferred). Each card plots you and the pick together.

More from this source