This moment, Alex's mile swim, it means everything to him and to all of us.
At Evernorth Specialty Services, we help providers make sure patient care doesn't get interrupted, especially for rare, chronic and complex conditions.
Now Alex and his care team have his hemophilia under control with life-saving medication and care management from Evernorth's specialty pharmacy Acredo.
So Alex can be the last one out of the pool.
Evernorth Specialty Services, because every moment counts.
Visit evernorth.com slash specialty to learn more.
What's up?
Adam Grant from Work Life, a TED podcast here, and I want to tell you about something exciting.
You're watching Every Dollar, and Walmart Business helps you stretch each one.
From office supplies to snacks and cleaning gear, you'll get everyday low prices, plus easy bulk ordering and fast delivery.
And with tools like spend tracking and multi-user accounts, staying organized is simple.
Save time, money, and hassle at business.walmart.com.
It's free to sign up.
Welcome back to The Nature Podcast.
This week how to heat gold beyond a theoretical limit and what researchers think about the threat of nuclear war.
I'm Lizzie Gibney.
And I'm Nick Petrichow.
How hot can gold get?
Well, the temperature solid gold can reach does have a theoretical limit, but a new study in Nature suggests that researchers might have shot right past it.
We were shocked how hot it was.
This is Tom White, a researcher who's interested in extreme states of matter like say, the state of gold if it were heated to incredible temperatures.
In this case, according to Tom, to 14 times its melting temperature, while remaining a solid.
But these extremities?
Research is pretty difficult and scientists often struggle with even relatively simple sounding things like working out how hot something is.
In my field, where I create these extreme states of matter, we haven't been able to measure the temperature of these states very accurately.
And this has been a long-standing challenge for many decades, measuring the temperature.
These states of matter we produce exist for like a nanosecond, a billionth of a second.
You can't take a thermometer and stick it in and say, what is the temperature of this material?
So Tom and a team of researchers have been developing a way to measure temperatures at the extremes they landed on using x-rays.
In fact, they used a Leinach coherent light source, a 3km long x-ray laser at the SLAC National Accelerator Laboratory in California, and they used it to deliver a short pulse of x-rays to a material, in this case some very hot gold.
The way this works is by measuring the energy of the x-rays after they hit the material.
All the x-rays have the same energy at the start, but they then scatter off the material and the energy gets shifted.
This shift, according to Tom, is related to the temperature of the material.
The thinking goes that as materials get hotter, their ions move around more, causing a bigger spread in the energy levels of the x-rays that hit them.
And while the principle is simple enough to physicists, it's very difficult to pull off in practice.
Even though it's the brightest x-ray source on the planet and it produces 10 to the 12, so that's a one with 12 zeros after it.
It produces that many x-rays we measure on our detector normally one or two x-rays per shot.
So to get an idea of the temperature we have to repeat the experiment multiple times at high repetition rate to get an average temperature out of all of the shots that we're doing.
So it's really pushing the technical limits of what we can achieve nowadays with our most advanced facilities.
The team tried the technique out on gold because it's heavy and so will scatter more x-rays.
To see how well the measurement worked, they heated up the gold with a short burst of energy from a laser and then used their x-ray technique to measure the temperature.
That's when they saw something surprising.
So we saw that we were able to heat this gold foil up to 20,000 Kelvin.
14 times the melt temperature, which is a lot higher than this predicted limit of three times the melt temperature.
So yeah, this was a total surprise to us.
We did not expect to see that.
Now, materials can go past their melting point and remain solid.
This is known as superheating.
An example of this you may have experienced is when you heat water in the microwave.
Heat it enough and it can remain a liquid even when it's gone well past its boiling point.
But then add a spoon or disturb it somehow and it explosively boils.
Don't try this one at home.
Even this superheating has limits though.
In a famous theory published in Nature in 1988, it was suggested that there is a critical point where liquid or solids can no longer remain in their current state.
It's dramatically called entropy catastrophe, and it refers to when the disorder or entropy in a solid material would meet that of its liquid counterpart.
Going beyond that point would break the second law of thermodynamics.
For gold, this entropy catastrophe would be at around 3000 Kelvin, or at three times its melting temperature.
I want to be really clear here that I am not proposing that I have broken the second law of thermodynamics.
That would really upend all of physics, okay?
So instead, what we found is that if you modify the equations to account for heating something up really rapidly so if you heat something up faster than it is allowed to expand then you can modify the equations that describe this limit.
And the two lines that describe the entropy of the solid and liquid.
They don't cross at any temperatures anymore.
So they just move away from each other.
And so if you do that and modify the equations, you find that actually there is apparently no limit, at least from this particular perspective.
There's no limit to superheating.
Not everyone is convinced that this is the case, though.
So I didn't believe they measured that temperature.
I thought it's not temperature.
That was my first thought, that they're measuring something else.
This is Artur Tam, a computational physicist who studies laser heating of materials and who's been writing a News & Views article on the new study.
We definitely see what they measure, but we don't know if that's the temperature or that's something else.
According to Arter, the uncertainty could be due to the nature of the very hot gold.
When a material is heated by laser like this, the dynamics of the individual ions that make it up might be quite different from more conventional heating.
Also, the excitation of the ions by the laser could affect the melting temperature.
Altogether.
This could mean that the temperature measurements that Tom and his colleagues got could be thrown off.
And Arter thinks that the measurements that Tom and his team show aren't directly the speed at which the ions in the material are moving.
So while they may relate to temperature, they also may not.
However, Arter does think this would be fairly easy to resolve in future studies.
We need really good models that would reproduce this sample or this experiment.
So the nice thing about this experiment is that this is something that we can exactly model in a computer, because it's such a well-defined problem.
We have a clean sample, relatively small, so we can reproduce that in our computer.
And from the simulation, we can exactly see what is each atom doing.
What's the velocities?
What are the positions?
Tom says he understands Arter's concerns, but he notes that the specific properties of gold should allow for an accurate measurement of temperature using his X-ray method.
Arter says that this experiment has produced useful data, regardless of what the temperature of the material might actually be.
It expands what researchers know about materials when they reach very high temperatures.
And Tom also thinks that this research could be useful for the design of materials that need to get really hot.
You know, we are designing a lot of interesting reactors and re-entrant spaceships and things like that, where these materials get subjected to extreme amounts of radiation.
And so understanding the stability limits of these materials how hot can they get before they break, and things like that is extremely interesting.
Studying other materials could also help validate the method used in this study.
And in the future.
Tom hopes that this work could help us understand what happens in the extreme conditions inside planets.
We just did an experiment.
Six night shifts.
So we were working from 6pm to 6am for six nights in a row.
And we were investigating, in this case, the behavior of iron.
And one of the differences instead of these very thin foils, we were using much thicker samples of iron that we were hitting with what's called a long pulse laser.
So we were squeezing the iron, not just heating it up, but squeezing it to create high density, high pressure conditions, to look for when iron melts at conditions that are found inside planets.
So we can more precisely identify the melt temperature of iron inside planets and describe how the interior planets might behave.
So that's where we're sort of taking this diagnostic now in that direction.
And it was extremely successful, if not totally exhausting.
That was Tom White from the University of Nevada in the US.
You also heard from Arta Tam from the University of Tartu in Estonia.
For more on that story, check out the show notes for some links.
Coming up with increasing political polarisation.
Researchers are warning about the threat of nuclear war.
Right now though, it's the Research Highlights with Dan Fox.
Island life has given the world's largest bats their enormous wingspans.
Flying foxes are a group of some 80 bat species found on islands across the Pacific and Indian Oceans.
And while they are generally large, some species have a wingspan as small as 0.3 metres.
To understand this size diversity, researchers analysed their evolutionary relationships through variations in their genes.
They found that island isolation and an evolutionary process known as character displacement were the main factors that led to this divergence in body size.
In this process, bat species would have adapted to have different wingspans, so they don't end up competing for the same resources.
The authors say these results emphasise that natural selection in island environments can produce unique physical traits.
Flap over to Proceedings of the Royal Society B to read that research in full.
A sugary diet early in life rewires the brains of mice and affects their learning as adults.
Excessive sugar consumption is linked to obesity and metabolic disorder in children, but also to alterations in motivation, learning and other behaviours.
To understand the possible neural basis for these changes, researchers fed mice with either a sugary solution or unsweetened water.
Compared to the rodents reared on unsweetened water, the sugar-fed mice exhibited altered brain activity and behaviour later in life.
Responses to sugar were blunted, learning-related activity emerged more slowly and connectivity between brain areas was reduced.
Both sets of rodents learnt an association between an odour and a guaranteed sugary reward.
But when the reward became less predictable, the sugar-reared mice were faster to learn that the rules had changed and quickly give up anticipating a treat on the basis of the odour cue.
The researchers suggest that these long-lasting effects in mice highlight the need for further research.
Read that paper in Current Biology.
Last week saw the 80th anniversary of the Trinity nuclear test, which ushered in an era of atomic weapons.
While the threat of a nuclear conflict might seem like a remnant of Cold War era history, many researchers are warning of the danger to the world now, in light of increasing polarisation and more nuclear-armed nations.
Reporter Alex Whitsey has been writing about some of the chief concerns that scientists have and what they're trying to do about it.
She spoke to Benjamin Thompson about the mood right now among researchers.
The mood is very much one of trying to raise public awareness, which sounds a bit boring, but for many decades the risk of nuclear war was a really big thing on a lot of people's minds, right.
It was enormous if you lived through hydrogen testing by the Soviet Union in the US in the 1950s.
It was huge if you grew up during the Cold War.
If you're a child of the 80s like me, you remember all sorts of nuclear doom threats all the time.
But today it's Maybe fading a bit from a lot of people's minds.
We have so many other pressing problems, especially climate change.
That's much more emergent in the minds, especially of folks who are under 40.
So the scientists I've been talking to, a lot of whom are nuclear security experts from way, way back, are very much like hey guys, this is not an old problem, this is a big, big problem and getting bigger all the time.
Right, because i think of the mutually assured destruction that we were all warned about in the 80s, where there was two big players which just about held the peace because neither side could win and both knew they couldn't win.
Things are different now.
Things are different.
And one of the main reasons things are different is there's a whole bunch more players, right?
So today, we've got nine countries that have nuclear weapons.
So not just the US and Russia, which has the Soviet Union's old stockpile.
We have many more players on the stage.
So it's not just two biggies kind of pointing their missiles at one another from thousands of miles away.
It's nation states that are having conflicts all the time.
They are threatening one another.
They might be firing conventional missiles, but you know they've got nuclear warheads in their back pocket.
And so that changes the dynamic entirely.
And with that in mind, there was a meeting last week in Chicago where researchers were essentially kicking the tires, so to speak, in the face of this new reality.
Yeah.
So the long history of top scientists warning of the dangers of nuclear war because, of course, top scientists brought the atomic bomb into existence.
So.
Right after Trinity, right after Hiroshima and Nagasaki and the bombings in World War II, you already had top scientists and Nobelists mourning of essentially this horrible thing that they had created and brought on humanity.
And then what happened in Chicago was kind of the latest of these gatherings.
So there were a number of nuclear specialists along with a number of Nobel laureates.
The idea being to kind of look at the threat today and talk about what are the scenarios in which nuclear war could happen.
How could we make it not happen?
How can we defuse these tensions?
How do we prevent nuclear war today?
And you've been looking at some of the things that they discussed and it seems like really, your article focuses on two things and two subjects we've talked about a lot on the podcast and indeed in the pages of Nature, namely misinformation and AI.
So let's maybe talk about the former.
First, misinformation and, as you describe it, it's cousin, disinformation.
Obviously it existed in a different sense in the 1950s, but of course now a lie can get half around the world before the truth ties its shoelaces, as the old saying goes.
Yeah, absolutely.
So we're just in an information reality where everything is so much faster.
And when you talk about the fog of war, having deep fakes and stuff like that feeding into whether or not your country is going to like unleash nuclear weapons and somebody else, it just absolutely sort of feeds that.
One of the examples that some of the experts have pointed out was that in May of this year, India and Pakistan had a conventional weapons conflict.
But the, As this was going on, news broadcasters even and social media channels in both countries were kind of flooding the airwaves with deepfake images and manipulated images and just incorrect images taken out of context, all having to do with oh look, we're doing great.
Here's a military triumph.
And obviously we have no idea if the military leaders in these countries are taking this into account.
One would like to think they have good information.
But again, just this public perception, this public frenzy, this kind of feeding on social media posts and that information is wrong, deliberately to try to gin up militarism and nationalism.
And when you do that and people are firing missiles and they've got nuclear weapons in their back pocket, that's just a whole situation.
We didn't have 10 20, 30 years ago at that scale.
And let's talk about AI, because in this instance there are sort of clear examples of where AI is being incorporated into military decision making.
And researchers are concerned about the spread of this.
I thought this was fascinating.
I didn't know a lot about this area until I started calling around for this story.
And in just the last couple of years, experts have been doing a lot of research into especially new types of AI, like reasoning models, things that take a lot of input and come out with a suggestion that looks kind of eerily like human intelligence, although it isn't.
That's kind of what's new here.
For decades, of course, people have used some sort of automated if-then flow in their military operations.
Like, if a country does this, then we're going to do that.
And they use computer systems to speed up that kind of stuff.
But now we've got these reasoning models.
And one of the scientists I talked to, Herblin at Stanford, who has thought really deeply about this, points out that some uses of AI in military operations, even in nuclear operations, are fine.
Spellcheck is basically an AI, and if you're spellchecking your document describing your military operations, that's probably fine.
But maybe there are ways it could help you.
So maybe you're a commander and you're looking at your options.
And maybe you want to ask that AI like of my military options which would violate the international laws of war, which would result in fewer casualties.
You know these military operations and especially in nuclear operations.
You've got page after page of option A B C D, E and all the scenarios of how things would work.
And AI models could feed into that.
And maybe that'll help a commander make a decision.
Of course, the other side of it is less appealing.
The other side of it is very less appealing.
So the joke being you don't want to give the launch codes to chat GPT.
Because what they call nuclear command and control like the finger on the button.
Who is the person who says you know what?
I'm going to launch this nuclear armed ICBM at my enemy and You do not want an AI in charge of making that decision?
So there's been a lot of discussion about quote unquote humans in the loop, right?
We talk about that in a lot of other AI issues, but you really, really want humans in the loop when you're talking about pushing the launch button.
That's kind of a huge area is like what do we know about how countries are incorporating AI tools into making launch decisions?
I guess it's hard for folk to get a sense of where that is happening though, because by its very nature these things are particularly secretive.
Nuclear launch operations are among the most classified of any state secrets for any nation.
So the ones that are sort of most transparent are not very transparent at all.
But we do know some.
Some countries have made statements about keeping humans in the loop.
Notably the US and China.
Last year both said that to some extent they would keep humans in charge of that life-altering, world-altering potential decision.
Other countries who have nuclear weapons have also indicated that yeah, they kind of think maybe it's a good idea to have the human there making the last, the final call.
But so much of this is obscure.
So we don't really know to what extent all countries are using AI in nuclear operations and how much they're thinking about it now.
I guess this secrecy means that researchers kind of have to discuss this in the abstract.
But more and more of these comings together are happening, where researchers are trying to figure out what's going on.
You write about a few of them as well.
Yeah, this is an absolutely new field.
Like again just in the last couple of years, because no one had really thought about this before.
Suddenly these incredibly powerful reasoning models showed up.
So you've got this world where you have scientists and policymakers and and government officials or people who represent various state governments having these strange conversations, where nobody knows what they mean when they talk about AI and what they mean when they talk about nuclear command and control.
And there needs to be a lot more clarity.
There needs to be a lot more discussion.
So in talking to experts, they're like we need people in the room together talking about what's meant by all these things.
So nobody gets confused and makes the wrong decision.
You've said there that science was central, of course, to the development of the atomic bomb.
Do you think governments and politicians will heed scientists and researchers?
It's one thing getting together and saying this is bad.
Do you think that'll be listened to?
What sense do you get from the people you've spoken to?
Yeah, I spent a lot of time pressing some of the folks I interviewed about this, because of course, it's one thing to say oh, nuclear weapons are bad, please don't use them.
But if no one's paying attention to that, that doesn't get us anywhere.
So a lot of the experts that I interviewed feel very strongly about how this needs to be much more on the public agenda.
So in the Cold War, there were massive demonstrations against nuclear weapons.
In West Germany there were like a million people in the streets or more demonstrating against nuclear weapons.
And this was a common feature through the Cold War, especially.
We don't see that now.
There are a lot of things competing on our plate.
So what the Nobelists and the scientists who were making this call really want is for folks to realize how severe the threat is, so that they can escalate it and make it a thing that people talk about in political debates when someone's running for leadership of a country, which it has not really been in the last five, 10 years.
Alex, obviously, this is an apocalyptic thing to write about.
How did this affect you with all this research?
Did you uncover things that you didn't know that really have maybe kept you up at night?
I will say that I read a lot of history of the Manhattan Project and I've been to Los Alamos and the Trinity site.
And actually my first science writing job was at Oak Ridge National Laboratory, which was part of the Manhattan Project.
So a lot of that kind of technical science US development, I had been steeped in a lot.
But honestly, in reporting this piece at the conference in Chicago, one of the representatives of the survivors of the bombings in Japan spoke and reading more as well about the accounts of the survivors and what it was like to experience that day.
That was very, very affecting to me.
Yeah, I think about it when I wake up at night.
That was Alex Whitsey speaking with Benjamin Thompson.
For more on that story check out the show notes for a link to Alex's feature.
Finally on the show, it's time for the briefing chat, where we discuss a couple of stories that have been featured in the Nature Briefing, Nature's daily roundup of science news.
And this time Lizzie, I believe you've got a story you've been writing that you wanted to tell me about.
That's right so another day another AI model.
Oh good um, what's cool about this model?
So it's called Kimi K2.
It's from a company called Moonlight AI and some people are calling it another deep seek moment.
So you might remember, we reported quite extensively on deep seek at the beginning of the year.
It was a open model and it came out of China and it was incredibly good and it caused a bit of a stir, and it seems like this Kimi K2 model might be doing something similar.
So people are excited about it because it's open, and the key thing about that is it's It's open weight, which means you can download and take the whole model and then you can fine tune it.
You can build upon it yourself.
So as a scientist, that can be something that's really important.
If you have a bunch of data that you want to incorporate, you can do that if it's an open model.
You can further train your model on that data.
And the other exciting thing is, yeah, it's coming out of China.
And that seems to be a bit of a pattern.
And what else can this one do?
Because as you mentioned, DeepSeq was also this open model.
Does this one have increased capabilities?
So first of all, yes, this beats DeepSeek.
It also beats Quen, which is another very popular open model that's out there.
It's the best open model that's out there.
Several people have now said.
The company has reported how it's performed on lots of different benchmarks.
And we know that there are always problems with benchmarks.
You know they can be gained and it depends which ones you pick.
But, especially for activities like coding, it just seems to perform extremely well maybe not the very best, depending on exactly which benchmark you use, you know compared to some of the proprietary models that are made by companies like OpenAI and Anthropic in the US.
But it's very, very good.
The other thing is, it's specifically designed to be agentic.
And what that means is, you But it can pull in, it can work with lots of other tools.
So you can imagine that that would be really useful if you want to search the Internet, if you want to use some specific other program.
What this can do is pull together all the different strands that you might need for your kind of longer term multitask project.
That's what it's specifically been trained to do.
Right, right.
So if I wanted to do some sort of mathematical simulation say, then I could be like hey, can you help me with this?
And it could use different software and that sort of thing.
In theory it would have the capability to know how to bring those things together, to make the right choices, to do that and just to kind of hold in its very air-quoting brain the project that it's doing and to do this kind of that multi-step activity.
So I wasn't able because it's only just come out.
I wasn't able to speak to many researchers who have really had some great successes with it yet.
So a lot of what we're saying is relying on the benchmarks that the company has put out there.
But there certainly seems to be quite a lot of excitement about it.
And there's one more thing that it does seemingly perfectly. particularly well, which is writing.
And this was something that I found fascinating.
Both Deep Seek and Kimmy K2 seem to write in a style that is quite different from the Western models.
And I think we all kind of know now we're starting to learn what it sounds like to have something written by AI.
It's quite, you know, there's quite a trademark style.
And these models don't have it.
And we're not really sure exactly why that is, if they've trained in a different way or on slightly different content.
But certainly Kimi K2 has topped all of the benchmarks for creative writing that are out there and for like, emotional intelligence.
So that's like how you would construct a role play and how people would act.
So, yeah, that's another little quirk to it.
And the other thing that caused a bit of a stir with DeepSeek is it was a model that was made very cheaply.
Is this the case for this one as well?
That is not very clear.
It certainly trained an awful lot and it is a very large model.
It's, I think, a trillion parameters, and parameters are these kind of like, almost like the weights between the different associations that the model makes?
So it's a big model.
The technical report actually just came out and I've not had a chance to go through it finally yet.
So we're not quite sure how cheap it was. to make but i i would take a guess and say they're probably some you know some efficiencies that have been made there there seems to be a lot of incentive to make these models more cheaply and just ability to make them more cheaply in china so that is definitely a possibility and you know i remember with deep seek that it caused a bit of a stir amongst ai researchers and companies and also on the stock market as well has there been a similar kind of reaction to this This has bubbled under the radar a little bit more.
So we haven't yet seen, certainly haven't seen any stock market plunges.
But there is definitely a sense that this is not going to be the last really good open model that we're likely to see coming out of China.
And there is a bit of disquiet among proponents of this kind of open source strategy.
This is not an open source model.
It's an open weight model.
I should say that there is a bit of a distinction.
We talked about that before, but it's very much within the open source community.
And that community in the United States is saying, like, look at what's happening.
All of these models are open in China.
And that means a lot of people are going to pick them up.
They're going to become the default in many areas of coding, in many areas of research, just for that reason that I mentioned the beginning.
They can be built upon and they are able to be studied as well, in a way that you can't study a proprietary model.
So there is a little bit of concern that the US is being left behind in that sense.
OpenAI has said that they will bring out an open model.
So I will look forward to seeing that when it happens.
But there is definitely a sense that the center of mass could change.
You know be shifting a little bit more to China, especially if you know these are just two companies and they're completely different companies.
There are a bunch of other startups that have gained huge traction in China that we, I'm sure, haven't heard about over here, but are doing great guns in China.
And, you know, so there may be many, many more models like this to come.
And China and the US have had a bit of a strained relationship recently.
And one of the things that happened was there was restrictions on what sort of technology could be imported to China from the US.
But that doesn't seem to have really slowed down their progress in AI, at least based on this.
That's right.
So what happened in the case of DeepSeek is they had used NVIDIA chips and NVIDIA is a US company, but they had used some of them that they were allowed, despite the export controls.
And it's very likely that a similar case has happened here.
But of course, a lot of the technology is also being made now over there in China.
And there are strong incentives to create their own kinds of chips.
And I don't think they're any better yet than the US equivalents, but that is probably the direction that we're also going in.
And you mentioned as well that obviously researchers can fiddle about with the weights and things.
They can sort of interrogate this model a bit more than maybe some others.
And perhaps it can use other software as well.
So do researchers think this might be a bit of a boon for science?
I think any model like this is going to be hugely beneficial.
Already kind of.
You know, the workhorse of the scientific community is ALAMA from META, or that's what it was for a long time, because it was open.
It was such a big deal.
That then got replaced probably by Quen, this other Chinese model that came out a little while ago, and then DeepSeek, and now this Kimi K2.
So I think it's very likely that this will become a go-to model for many researchers.
Fascinating stuff.
I'll be interested to see what else comes out in this sort of open AI model space.
But I wanted to go on a slightly different tack for my story.
So I've been reading about dolphins this week and how some dolphins use sponges as a way to hunt.
Wow, okay.
How does a sponge help you hunt?
Well, so this was a story I was reading in the Associated Press, based on a paper from the Royal Society Open Science Journal.
And what they're doing here is they're using the sponges to essentially protect their noses from these rough stones that are at the bottom of the seabed.
So this is in Shark Bay in Australia, and some dolphins essentially just wear the sponges almost like a little hat and then use it to dig through the stones at the bottom to then disturb fish that are camouflaged and then catch them.
Oh, okay.
So that's kind of like protection for their nose while they're digging and flushing out all of the yummy things to eat.
Yeah, exactly.
And this has been a thing that's been known for a while that these dolphins do.
But the question researchers have had is what effect this has on their other abilities, like echolocation.
So you know that echolocation is when dolphins use like click sounds and that sort of thing to navigate and to find their prey.
But obviously, if you have a big sponge on your nose, that might get in the way of it, and so researchers used an underwater microphone to determine what effect this might be having.
And obviously we can't know what's going on inside the dolphin's brain, but for sure the sponges seem to affect the, the sound that they produce with echolocation, and so the researchers are proposing they're having to adapt to this and learn how to do this in a way, and they said it would be like hunting using a blindfold.
And so they have this muffled echolocation.
How can they still?
Does it just make?
It makes it just worse, like they, they aren't as accurate with their echolocation, or what.
How does it affect it?
Yeah, the researchers likened it to wearing glasses with the wrong prescription.
So if anyone wears glasses, they know what it's like to wear glasses with the wrong prescription or just not wear them.
So everything's a little bit blurry.
It's hard to see.
But if you put the time and effort in, you could probably figure stuff out.
You could probably, OK, that blurry thing is, you know, a pillar.
I should avoid that, that sort of thing.
And so they think, over many years they learn how to deal with, I guess, this sort of blurrier world that they're now interacting with.
Oh, wow.
So there must be real benefits from this kind of method for hunting.
Yeah, that's what the researchers believe.
But what's interesting is that only about 30 dolphins actually do this.
So about 5% of the overall population of these particular kinds of dolphin.
And so, because only a few dolphins do this, researchers were wondering why it was that only these dolphins have bothered to do it.
And it seems that it's because it takes a very long time to learn.
This is essentially muffling their echolocation and, you know, making it much more difficult.
As i say, it's almost like hunting with a blindfold for the dolphins, and so they learn from their mothers, and they learn over a period of years, and many dolphins don't stick with it as well.
They don't bother learning this all the way through.
It sounds like a phd.
Yeah, exactly.
So it's a really hard skill to learn.
So while it may offer benefits, the difficulty of it may have limited it just to this population.
But it's an interesting insight into some of the difficulties of how animals learn to use tools.
And the kind of social dynamics of being a dolphin, right?
That you get these skills can be passed on from the parent to the child.
But then, even if they seem to have some great advantage when it comes to hunting, you need to learn the whole package.
And that takes a long time.
It's quite hard to do.
So you can't just watch another dolphin do it and then do it yourself.
It's fascinating.
Yeah, no, it takes a lot to do.
And it's another example of this sort of cultural transmission of skills.
And this is something that for a long time researchers believed only humans do.
Because, you know, we go to school, we learn things, we learn from our parents, that sort of thing.
That is one of the great advantages we have over other animals.
But this is another example of other animals doing it.
And it may be quite common in the animal kingdom also.
But yeah, this is an interesting sort of insight into the costs and benefits of learning a particular skill for these dolphins.
Fascinating.
Thank you, Nick.
And if those stories have intrigued you listening at home, check out the show notes for links to them and for a link of where you can sign up to the Nature Briefing to get more like them delivered directly to your inbox.
And that's all for this week.
If you've enjoyed listening, let us know.
You can leave us a review on your podcast app of choice or you can send us an email to podcast at naturecom.
Or we're on X and Blue Sky with the handle at nature podcast.
I'm Nick Petrichow.
And I'm Lizzie Gibney.
Thanks for listening.
This moment, Alex's mile swim, it means everything to him and to all of us.
At Evernorth Specialty Services, we help providers make sure patient care doesn't get interrupted, especially for rare, chronic and complex conditions.
Now Alex and his care team have his hemophilia under control with life-saving medication and care management from Evernorth's specialty pharmacy Acredo.
So Alex can be the last one out of the pool.
Evernorth Specialty Services, because every moment counts.
Visit evernorth.com slash specialty to learn more.
Skip the gym, not the workout.
With Hydro, you get a full body workout in just 20 minutes, hitting 86 of your muscles in one smooth, low impact motion.
And now there's the new Hydro Arc, Hydro's most advanced rower yet.
Arc introduces hydrometrics tracking your power, endurance and precision every time you row, so you can actually see your progress.
Go to Hydro.com and use code ROW to get $100 off any Hydro rower, including ARK.
That's H-Y-D-R-O-W.com, code ROW.
