An officer and a gen AI: the future of war

The Economist. Hello and welcome to The Intelligence from The Economist. I'm Jason Palmer.

And I'm Rosie Bloor. Every weekday we provide a fresh perspective on the events shaping your world.

The DNA sequences of lots of long-dead beasts are already known. What scientists hadn't discovered is how that DNA is folded and tucked into chromosomes, the real owner's manual for how to run, say, a woolly mammoth. That, though, just changed.

And everyone knows the painting The Scream by Edvard Munch. But few were haunted by it like Pål Enger was. So he stole it. Our obituaries editor describes a thief who, under different circumstances, might have been a friend of the artist. First up though.

The war in Europe had—

Battlefields have long served as proving grounds for new technologies.

To break the stalemate, Britain had introduced the tank. Germany was using liquid fire.

And those technologies have in turn changed the character of war, from tanks to radar and the awesome destructive power of the bomb.

The force from which the sun draws its power has been loosed against those who brought war to the—

Computers too have played their part. Colossus was built in 1944 to crack Nazi codes, and by the 1950s, a digital control and command system was organizing America's air defenses. In the decades that followed, the role of machine intelligence grew but remained quite small. Soon it will become pivotal, and it could transform the nature of warfare altogether.

Artificial intelligence is trickling into almost every aspect of war.

Shashank Joshi is The Economist's defense editor.

That is prompting some tricky questions about the nature of war itself and the human role in it. I think a really important gulf is opening up between the lawyers and the ethicists who worry about the role of AI and armed forces who see enormous opportunities and fear being left behind by their rivals.

You say AI is trickling into every area of war today. Tell us about some of the cutting-edge technology that we're seeing.

So start with the really boring stuff at one end that's dull but essential. That could be logistics, that could be maintenance, that could be HR. AI is being used in all of that. So for example, using AI to predict when a particular unit on the front may need more shells. And sending it to them before they even ask for it. At the other extreme, Rosie, the bit that people tend to be most interested in, perhaps most exercised by, is the killer drone bit, the weapons that are fitted with some kind of AI. Maybe that's computer vision that can recognize a tank, maybe that's the ability to maneuver across a battlefield and maneuver through the inside of a house using autonomous navigation. And that's the sort of sharp end. I think the really interesting bit is everything between that and that the line between intelligence, which is about spotting patterns in data, and command and control, which is about doing something based on that intelligence, is blurring.

So give me some examples of these bits in between that are making militaries function.

I think one really interesting example of this is the Lavender system being used by the Israel Defense Forces, the IDF, in Gaza. This has had a lot of attention recently. The claim is that this is a kind of AI system that's being used to identify tens of thousands of Palestinians as targets with human operators only giving very cursory scrutiny before ordering the strike. Now, in practice, that is likely to be a version of what we call a decision support system or DSS. And what's effectively happening is that computers are not just processing lots of data, they're bringing recommendations or options to commanders. And at some point, humans can treat those as a kind of menu that they pick and choose, or they can treat those as something to simply accept. And at that point, the machine is almost co-commanding with the human. And I think that is the kind of future that many armies see.

So spotting is okay, but decision-making, recommending is obviously very much getting towards the killer robot territory of science fiction. Tell us a bit more about the kind of ethical questions this is all prompting?

On the one hand, Rosie, I could give you examples from the Vietnam War in which America was pulling together data from sensors, they were processing them in big IBM mainframe computers in Thailand, and sending the data to planes. And you can argue that is also an example of kind of algorithmic computer-enabled warfare. What's the difference? The difference, I think, is that today's software enjoys much more computing power, it has many more sophisticated algorithms, more data than ever. And I think that is what is changing the ethical discussion, partly because if your computer system is giving you the option of 1,000 targets, then it really isn't that easy for any human or any team of humans to sort through that in a reasonable period of time. And if the opposing army is saying the computer can decide which targets are hit, but you're relying on humans for that, I think the fear is that actually this is going to slow you down and allow the opposing army to get the advantage. But if you simply accept the recommendations given by the computer, then of course, you are going to make mistakes, you're going to see the algorithm make some catastrophic errors, and there may be areas where computers are simply unable to do their job properly. So for example, can they really tell the difference between a child holding a toy gun and a soldier holding a real one. And those problems are really going to get worse and worse as these command systems are used more and more.

So I think what you're saying is that we've always asked computers to help us make decisions, but obviously these decisions are much bigger. We're now asking computers to do things that humans can't while being terrified that computers can't do things that humans can. Is that correct?

I think you summed it up very well. And the fear is that not only are we asking computers to do a lot, that as these algorithms become more advanced, we understand them a bit less, but also they're more unpredictable. They may be trained on data from Ukraine, but then if you're applying them in the case of Taiwan, they might result in very unpredictable outputs. The response to that from military types is to say, all that may be true, but war is already very messy. Human intelligence Police officers who have to sort through targets are subject to exhaustion, they're subject to mistakes themselves. And actually, the baseline isn't whether the computer is perfect or not, it's whether it is better than a human team doing the same thing. And actually, the ethicists and the lawyers, sometimes they have too rosy a view of conflict today, the way it's waged by humans, and so they expect too much of the machines. And I think that debate really is still playing out between those who are worried by the responsibility being imbued into machines and those who think actually that perhaps if the machines are more reliable, more predictable, not subject to emotion or fatigue, maybe they would even make more ethical decisions on some of these matters.

So is war in practice going to wait for us to have these debates?

I think to answer that, look at Ukraine. Ukraine and Russia are both putting AI in their drones They're using AI in their command systems to, to decide where artillery should strike. They're using it to find Russian generals. They're not waiting for these legal and ethical debates to resolve themselves. They're fighting in Ukraine's case for their survival. And I think hanging over this, Rosie, is also the kind of specter of a war involving big powers. NATO countries know they might one day have to fight Russia, an army that has now accumulated years of experience in putting AI into its systems. And has very few ethical compunctions about the effects of that. China is pursuing many of the same technologies as well. And so American commanders are asking, what's the People's Liberation Army gonna do?

Who's got the upper hand in the AI defense supremacy?

A few years ago, you would've found that Americans were fretting. China's pulling ahead of us in AI. They have vastly more data collected from things like surveillance cameras. They have fewer rules about using it. They have control over private companies. I think those concerns have abated a little bit. We've seen America pull ahead in cutting-edge foundation models, things like the GPT models inside ChatGPT. We've seen American chip export restrictions really savage some Chinese companies. And I think we're also seeing American companies test some of their products in Ukraine pretty successfully, showing that they can also build very impressive things. However, there is a worry, I think, among a lot of people that armies and armed forces in the West are still quite slow to evolve. The Pentagon still spends less than 1% of its budget on software. And I think there is a, a real sense that we can see the future of war in the distance. AI is an integral part of that, But the gap from here to there of really transforming today's armed forces remains really, really formidable.

Terrifying stuff. Thank you so much, Shashank.

Thanks very much, Rosie.

Artificial intelligence isn't just changing the face of war, it's also transforming espionage. On this week's episode of Babbage, our science and technology podcast, Shashank takes us into the shadows to find out how spies are using tech to gain an advantage. You can listen to Babbage by subscribing to Economist Podcast Plus.

Scientists can learn quite a bit about extinct animals from the hints they leave behind. Bones reveal body shapes and musculatures. Footprints hint at posture. Stature, weight, even running speed. Some dinosaur fossils contain stunning detail, like patterned, colored feathers. Maybe it doesn't sound so surprising. After all, moviegoers long ago became used to the idea that dino leftovers could be fashioned into tourist attractions.

There it is.

Welcome to Jurassic Park.

We've made Living biological attractions.

Back in real life, there's still a tremendous amount about past beasts tucked inside their genetic codes that paleontologists are still desperate to know. And some of that can be learned when the leftovers are preserved in a particular way.

For the very first time, scientists have found what they're calling fossil chromosomes.

Rory Galloway is a senior producer on The Intelligence and also reports on science for The Economist.

They're inside the cells of a mammoth that was frozen for 52,000 years in Siberia. And it might just change the way that we understand ancient life all over the planet.

Now, I'm a little unclear here because I thought that I had heard that DNA had been pulled out of a mammoth somehow before.

So you're right, Jason. We've had the DNA of a mammoth for a while, but it's normally found in these tiny little fragments of DNA. What they do is something called paleogenomics, where they study those fragments and they sort of line it up into the right positions based on the DNA of an elephant and sort of checking across. But what this loses is all of the complex folding and other structures that are present in the nucleus of the cell. So these chromosomes and other structures that are really essential for understanding exactly which genes are switched on and not in the cell. And to understand this a bit better, I spoke to Dr. Erez Aydin from Baylor College of Medicine in Houston. And he explained how different this was from what we had before.

Imagine DNA is like a ladder. What's happening in ancient DNA is the ladder is getting, like, sawed every 30 rungs, 50 rungs, 100 rungs. But when we're reporting fossil chromosomes, what we're saying is those little sub-ladders that get produced, all those little fragments of ancient DNA are still arranged in 3D. As they were 52,000 years ago in the living organism.

So for anyone who doesn't know, chromosomes are made up of extremely long chains of DNA, all squished down and packaged, and they help to structure that DNA so that it can be copied or put in the right places so that the right genes can be activated when they need to be.

Okay, and so how do these packets of squiggly stuff end up as what you're calling fossil chromosomes?

Well, it's not dissimilar to the process that makes beef jerky, just a little bit colder. So when flesh dehydrates really, really quickly, the nucleus of the cell with all of its genetic structure dries out really fast too. And in that process, what happens is that the chromosomes within that nucleus are preserved into this sturdy solid. In a technical sense, it's a glass, so a non-crystalline structure. And Dr. Aydin and his team call that solid chromoglass.

Two things help arrest molecular movement. One is cooling. The other is removing the water so they all get jammed in there in that molecular traffic jam. And that combination is what creates chromoglass.

But the team wanted to find this preservation happening naturally.

I started to test, you know, my dinner from the Thanksgiving invitation at Eris's place.

Cynthia Perez Estrada is also working at Baylor, and she was one of the lead authors on the paper just published in the journal Cell.

All sorts of other critters that I could find in the road on my way to work. Crickets, roadkill, mice, and even my bag. Just, you know, throwing different ideas on how to make it work.

Even though these samples were such poor quality, they still contained chromoglass.

The majority of the experiments failed, but there were a few that actually gave us the idea that this search was not— maybe it was not a waste of time.

Dr. Pérez Estrada's experiments suggested it would be possible to find chromoglass in ancient specimens of extinct species, provided they were kept dry. So she started looking for it in something a little bit older.

Which is how we come back to the mammoth, I guess.

It is. She was searching for 9 years and eventually was sent a sample— well, 2 samples in fact— from mammoths found in caves in northern Siberia. Both were females. One had been frozen for 39,000 years and the other for 52,000 years. And they were so well preserved that their hair, skin follicles, and even individual cells could be seen upon magnification. And nestled deep inside those skin cells was the naturally formed chromoglass, or these fossil chromosomes.

Okay, so once you've found it though, how do you get it to tell you all the things that chromosomes can tell you that DNA fragments can't?

Well, that involved an awful lot of complicated computer modeling and collaboration with a team at the Center for Genomic Regulation in Barcelona. And I don't know enormous amounts about that, but what's more important is that it worked. We now know for the very first time that mammoths had 28 pairs of chromosomes, which was exactly the same number as other elephants have, both species, the Asian elephant and the African elephant. And a group of international researchers compared this mammoth skin cell data with that of their living elephant relatives in other ways too. And found what's called epigenetic changes. So if the genome tells you how to build the beast, the epigenetic changes tell you how to run the beast. They tell you which genes are switched on and which aren't in its specific environment. And that's really exciting. Some of these genes were involved with coping with cold weather, perhaps unsurprisingly, and that included a gene for really long hair in the mammoth. But what's more exciting to me is what this means for species outside of just the mammoth.

Such as?

Well, Dr. Aidan expects chromoglass to be found in far more places now we know what to look for. And he also found that when this dehydration process forms this chromoglass, at the point that it's formed, it becomes seriously durable. It's really solid stuff. And in the paper, he and his colleagues theorise that it has a pretty extraordinary shelf life.

We estimate about 530 million years. Which is a very long time.

Now, assuming there is extremely good preservation, 530 million years takes you back way beyond the dinosaurs. That means we could find DNA from the Cambrian era. That's a time before any animals had crawled out of the water and onto land. That's how far back we're talking. But as I was discussing this with Dr. Aidan, he had an even more fun way to think about this.

I wouldn't take it like too seriously, but you could imagine there are cold, dry environments on comets. At the rate that comets move over 530 million years, material from Earth would have arrived at most of the Milky Way aboard a comet. So that's really robust.

Now there's another fun way to think about this, Jason, and again, this is super speculative, but space is pretty darn cold and a much better environment for preserving this chromoglass-like substance. There was a massive asteroid impact at the end of the time of the dinosaurs that blasted rocks, debris, and potentially unsuspecting T-Rexes out into space. So I like to think about the fact that sometime in the future, astronauts might be looking at some asteroids floating about somewhere out there in space, and they may just discover some T-Rex jerky complete with chromoglass inside it, just waiting to give up the secrets of the dinosaurs.

That is bonkers, Rory.

Look, I'm not saying Jurassic Park is close, but it's definitely closer than it was yesterday.

Rory, thanks very much for your time.

Thanks, Jason.

There are some paintings that just provoke a strong reaction.

Anne Rowe is The Economist's obituaries editor.

Pablo Picasso's Guernica had red paint sprayed on it. The Rokeby Venus of Velázquez was attacked with a meat cleaver by a suffragette. And as for Edvard Munch's The Scream, that famous painting of swirling reds and blues with anguished figure at the center, that one, at least to Peyer Enger, seemed to be something that had to be removed totally from where it was. He was never exactly sure why he stole that painting, but one moment his ladder was up against the wall of the National Gallery in Oslo, and the next minute He was bundling this painting, which was then worth $55 million, behind the front seat of his car and driving around Oslo for hours, wondering what on earth he was going to do next. He said it was a magical, electrical feeling to have this painting with him. He was so glad to possess it. On the other hand, what now? He'd been planning this theft for a number of years. He tried to steal The Scream before. He'd put up a ladder against the wall of the gallery and he'd got, instead of The Scream, because he broke through the wrong window, he found Munch's painting called Love and Pain, which shows a red-haired woman ravenously embracing a man. He was disappointed for days that he hadn't managed to get The Scream. But on the other hand, he was very happy to put Love and Pain up on his wall and to make quite an exhibition of it before the police got word that he had it. He went to prison for 6 years for doing that. But in 1994, he was well aware that Norway was about to host the Winter Olympics in Lillehammer, that the police would be very distracted that day, and it might be an excellent day to try to do what he really wanted to do, which was steal the screen. It was not something you'd expect from someone who'd been brought up in a very poor part of Oslo and had got into gangs pretty early. But he had a rather busy and extraordinary life as a teenager. On the one hand, he was determined to be a very successful criminal, and he was. He made such a lot of money from stealing jewelry and profiteering that even at school, he was going around in flashy clothes with large watches. And a bit later, he was seen with the most luxurious cars, BMWs and Mercedes, and as he liked to boast, the only Porsche in Oslo. Which the rich folk would come round to the slum just to see him washing it. He had also been a very promising footballer when he was young. He'd played for Vårenga, which is a team in the Norwegian Premier League, and had a lot of talent. But he couldn't do both the crime and the football. And in the end, because the football didn't pay very much, Munch and crime did. He just let the football slide. He found, though, surprisingly, that he did rather like the paintings of Munch. He'd first encountered The Scream on a trip in primary school, so he was very young. He was so struck by the painting that it haunted his dreams for a long time, because it seemed to him that this anguished central figure was expressing what he felt when he thought of his stepfather. He was a violent man. And so he began to think about the painting more and more. And as he became an adult, he read a lot about it, and he would go to the National Gallery at least twice a week and just stand there looking at it. Of course, he would also notice if he visited that frequently that The painting wasn't glazed and it wasn't guarded particularly, and it was also right beside a window. And to someone with a criminal mind, that was interesting. Having done that, he found he'd earned himself 7 years in prison, but it had also made him a great celebrity because The Scream was a national treasure in Norway. There'd been consternation when it was taken. And when it was discovered that the man who'd taken it was also a footballer, there was some romance that clung to Palle Inge. And he added to his fame by escaping from jail when he was there. He waited until all the prisoners were taken on a field trip and ran away. And then disguised himself in a blonde wig and sunglasses and tried to get a train to Copenhagen. At the same time, however, his criminal career was not going terribly well. He did still manage to steal numbers of smaller artworks and jewels, but he was either tiring or he was losing his touch. And it seemed that by 2012, he was reduced to stealing taxi license plates and pairs of socks. But while he'd been in prison, he had taken up painting. At first he painted animals, and then he had started to paint versions of The Scream. But then he preferred to do abstracts of his own, and it was quite strange how painting after painting seemed to resolve into the swirling reds and blues of The Scream itself. And into that ghostly figure at the center with its hands in anguish clasping its head.

Anne Rowe on Paya Enger, who's died aged 57. Producers are Henrietta McFarlane and Kunal Patel, with extra production help this week from Emily Elias. We'll all see you back here tomorrow for the Weekend Intelligence.