Curiosity Daily

Talking to Whales, Pedestrian Patterns, Women’s Competitiveness

Episode Summary

Learn how AI may learn to talk to whales; why pedestrians don’t follow the shortest route; and competitiveness in women.

Episode Notes

Learn how AI may learn to talk to whales; why pedestrians don’t follow the shortest route; and competitiveness in women.

Researchers are using AI to understand whale clicks — and talk back to them by Briana Brownell 

Pedestrians are wired to follow the "pointiest" route, not the shortest by Cameron Duke 

Women are just as competitive as men, they just show it differently by Steffie Drucker

Follow Curiosity Daily on your favorite podcast app to get smarter with Cody Gough and Ashley Hamer — for free! Still curious? Get exclusive science shows, nature documentaries, and more real-life entertainment on discovery+! Go to https://discoveryplus.com/curiosity to start your 7-day free trial. discovery+ is currently only available for US subscribers.

Episode Transcription

CODY: Hi! You’re about to get smarter in just a few minutes with Curiosity Daily from Discovery. I’m Cody Gough.

ASHLEY: And I’m Ashley Hamer. Today, you’ll learn about how researchers are using AI to understand — and maybe even talk back to — whales; why pedestrians aren’t wired to follow the shortest route; and why women are just as competitive as men — they just show it differently.

CODY: Let’s satisfy some curiosity.

Researchers are using AI to understand whale clicks — and talk back to them by Briana Brownell (Cody)

Have you ever wanted to have a conversation with a whale? Well, some day, you might be able to, thanks to a new project called the Cetacean Translation Initiative — aka Project CETI.

It’s an ambitious project. The team is hoping to use artificial intelligence to model and eventually translate sounds made by sperm whales into human language. And maybe even to have a back-and-forth conversation with them.

Why sperm whales? Well, sperm whales are smart, and their conversations lend themselves well to analysis. Sperm whales communicate using what are called “codas.” A coda is a brief series of clicks that sounds a little bit like morse code. Sperm whales talk back and forth to each other using these codas, and their relatively complex structure suggests that the language is complex too. 

Another bonus: since these whales communicate across long distances, their communication must be mostly, if not completely acoustic. That means the data we need to collect is straightforward  — all we need is the recording of these vocalizations, rather than more subtle cues like facial expressions or gestures.

But the project needs data. In order to create a language model, they need to collect enough recordings for the language processing system to learn — and eventually, predict — what sound might come next. The research team plans to collect and analyze millions of whale codas.

Scientists are modeling the AI’s learning method based on how human children learn to communicate.  From birth, children hear language spoken around them.  They deduce the meaning of words from the context and content statistically until they eventually learn the language. 

AI language models can do this too. The result is a method to create plausible replies, even without knowing the meaning or grammar rules of the language.

This poses another problem. Even if the bot could talk to the whales, it’s tough to figure out the meaning of what it’s saying. For instance, if the bot learned to answer “How are you?” with “Fine, and you?” it could successfully carry on a conversation, but we humans still wouldn’t know what the conversation was about.  Scientists hope that the contextual clues will help them figure out the meaning of the codas and map them onto human language so we could translate back and forth.

Of course, it might turn out that whale conversations are really not that complex. They could just be saying the equivalent of “hello” back and forth, or telling each other about the weather where they’re swimming.

But even if it’s a boring conversation, it’ll still be pretty groundbreaking. And maybe the whales will teach us some new tales.

Pedestrians are wired to follow the "pointiest" route, not the shortest by Cameron Duke (Ashley)

The shortest path between two points is a straight line, but new research shows that most of the time, we humans only think we’re taking the shortest path between two points. Instead, we usually do something pretty different.

 

That is, we tend to navigate by using a technique called vector navigation. Instead of taking the most direct path, humans navigate in a way that allows them to spend more time facing their goal. That usually means taking more meandering paths through city streets as opposed to ones that might get us there the fastest.

 

MIT professor Carlo Ratti says he became interested in studying human navigation two decades ago as a graduate student. He realized that he would walk one path to his office every day, then take a different one back home. This is obviously inefficient, because there should be one optimal path between two points. Either one or both journeys were much more meandering than necessary. If he wasn’t taking the ideal path, he wondered what logic he was using to navigate — and what logic everyone else was using. 

 

As luck would have it, a few years ago Ratti’s lab scored a set of anonymized GPS data that could help him answer his question. In the dataset were 550,000 pedestrian journeys taken by roughly 14,000 people in Boston and San Francisco. Using this dataset, Ratti and his colleagues were able to chart real life pedestrian journeys on maps of city streets and compare those journeys to the most optimal paths between the same two points. 

 

They found that humans, on their own, are not particularly efficient navigators. Instead of following the shortest path, most of the journeys followed what Ratti calls the “pointiest” path. For example, pedestrians will almost always start walking the path that points most directly at their destination, even in cases where the shortest path might actually involve heading in the opposite direction to get to a more direct street.

 

This kind of “vector navigation” means we travel in ways that keep us faced toward our destination at the expense of efficiency. It might be unfair to call this inefficient, though. Other mammals like rats, bats, and howler monkeys navigate in similar ways, and that suggests that vector navigation might be the most mentally efficient way to travel. It might free up brainpower for other cognitive tasks like avoiding predators — or reckless drivers. If this is true, it means that we make our feet work harder so our brains don’t have to.

Women are just as competitive as men, they just show it differently by Steffie Drucker (Cody)

The gender wage gap is nothing new, but explanations for why it exists have come and gone. One newer theory is that women are less competitive than men are. That’s a hypothesis that’s easy to study — and scientists at the University of Arizona recently did just that. 

 

For the study, they engaged 238 participants in a three-round math competition — about half men and half women. Each participant was assigned to a four-member team. The first round was the same for every team: they had to look at a list of 12 numbers with decimals and pick out which two added up to 10. They were given up to 20 lists and every teammate earned $2 for each problem they completed within 20 minutes.

 

Things got more competitive in round two. In half the teams, the two participants who solved the most problems got $4 for each one while the other two teammates got nothing — kind of a “winner take all” model. In the other teams, the top two participants also earned $4 apiece but could decide to share some of the money with a lower-performing teammate. More of a prize-sharing scheme.

 

In the third round, the participants got to choose which payment plan they wanted. For half of them, that was a choice between a guaranteed $2 per problem or the more competitive winner-take-all model. The other teams also got the choice of the guaranteed $2 per problem or the more competitive prize-sharing model. 

 

The choices the participants made were different for men than they were for women — but not in the way you might think. Roughly half of the men in each of the two groups opted for the more competitive option — it didn’t matter whether it involved a winner-take-all or a prize-sharing model. But the female participants strongly favored the more competitive option when they had the chance to share their winnings. A whopping 60 percent chose that option when it involved the prize-sharing model, while only 35 percent went that route when it was winner-take-all.

 

The researchers aren’t sure why this is. Did women prefer this model because it let them smooth over bad feelings with losing teammates? Or did they like having control over how rewards are doled out? The team is already developing a new experiment to find out. Either way, that concern for others sounds like an excellent trait in a CEO.

RECAP

Let’s do a quick recap of what we learned today

  1. ASHLEY: Scientists are using AI to understand the language of sperm whales — and maybe even talk back to them one day. Sperm whales communicate in a series of morse-code-like clicks called “codas,” and their complex structure suggests that their language is complex, too. The AI will learn their language the same way a child learns their native language, but scientists will have to work harder to actually understand the meaning of that language. Once they do, though, it could change our entire outlook about life on our planet.
    1. CODY: Although we know that animals communicate using sounds, there is a debate about whether animals use language. What’s the difference? Languages have three qualities that make them distinct from other kinds of vocalizations. First, languages have semantics. In other words, the meaning of a specific sound or series of sounds must remain the same.  Second, grammar. Languages have rules for how sentences should be formed.  Finally, learning. Innate methods of communication, like yawning or frowning might communicate, but they are not part of language. 
  2. CODY: Humans don’t take the shortest route to get from point A to point B. Instead, MIT research finds that we take the “pointiest” route — that is, we take the route that keeps us facing our destination for as much time as possible. Other animals do this too, which suggests that it might be inefficient from a travel standpoint, it might be more efficient for our brains.
  3. ASHLEY: Women are just as competitive as men — they just show it differently. That’s according to a study that found that women opted to be more competitive than men if they were given the option to share their winnings, but less competitive if they had to take all the winnings for themselves. The researchers aren’t sure why that is, but they’re already designing a new experiment to find out.

[ad lib optional] 

CODY: Today’s writers were Briana Brownell, Cameron Duke, and Steffie Drucker. 

ASHLEY: Curiosity Daily is distributed by Discovery.

CODY: [AD LIB SOMETHING FUNNY] join us again next time to learn something new in just a few minutes.

ASHLEY: And until then, stay curious!