Indus River in Kashmir crossfire, and gene-stealing cancers
In this edition of The Naked Scientists: Will - and could - India switch off the water supply to Pakistan amid mounting tensions over Kashmir? Also, how a transmissible canine cancer from 8000 years ago is shedding fresh light on tumour biology - and how cancers can steal foreign genes, today. And, testing out a virtual reality tool designed to help us confront our fears of speaking in public...
In this episode
01:10 - India threatens to deprive Pakistan of Indus River's water
India threatens to deprive Pakistan of Indus River's water
Sajjan Gohel, Asia-Pacific Foundation & NATO DEEP’s Global Threats Advisory Group
The Indus River sustains millions of people across India and Pakistan, weaving through one of the world’s most volatile regions. In 1960, the Indus Waters Treaty - an extraordinary agreement between these hostile neighbours - divided its waters. And the treaty has survived wars and diplomatic standoffs in the decades that have followed. But today, that fragile accord faces mounting threats. Recent terror attacks in Kashmir have reignited tensions; and the Indian leadership have hinted that, by way of reprisal, restricting the water flow to Pakistan could be on the cards. As dam projects accelerate and trust deteriorates, water is no longer just a resource: it's becoming a political weapon. Pakistan fears that, one day, the taps could be turned off entirely. We asked Sajjan Gohel, International Security Director at the Asia-Pacific Foundation and Chairman of NATO DEEP’s Global Threats Advisory Group, to explain what has put water at the heart of this heated dispute between India and Pakistan, and whether it really is scientifically feasible to cut off a whole country from such a massive water supply…
Sajjan - The Indus Water Treaty arrangement consists of six rivers. You have the eastern tributaries of the Indus, the Ravi, Beas and Sutlej and then you also have the western tributaries, the Jhelum and Chenab. All of them are located in the Himalayan region of Jammu and Kashmir in the north of South Asia.
Chris - And who controls the terrain that they originate from and flow through?
Sajjan - While most of the Indus river system originated in India. They flow into Pakistan so that gives India the upstream control which has often at times caused tensions in the past. Both countries realise that potential prolonged disputes over the water could escalate into a war and what it could also mean is that tensions would always play out.
Chris - And how when that treaty came into being was the water divvied up? And is there a sort of a share? So one country is allowed a certain proportion, the other countries allowed the rest or is it just that they agree that someone's not going to dam the whole thing?
Sajjan - The treaty governs how India is allowed to use or restrict water especially from the western rivers that are primarily used by Pakistan. It's designed to ensure the free flow of water and India ultimately does control the flow of it. It is an example of how it has actually continued despite various conflicts that India and Pakistan has fought going back from the wars in 1965, 1971, 1999 and yet the Indus water treaty still continues to exist and the waters continue to flow.
Chris - In recent weeks though we've seen tensions escalate and now India is saying they may shut off the water supply. Is that actually feasible though? Have India got anywhere to put that much water? Because it's not trivial if you have that water flowing off the Himalaya, you can't just say we'll return the tap off.
Sajjan - There is potential for India to do some activity that could cause problems for Pakistan. Track back to 1948, just months after partition, India temporarily stopped water flow to some Pakistani canals from the Ferozepur Headworks. So that heightened tensions and demonstrated the urgent need for there to be some kind of formal agreement. But you're absolutely right, it's not like there's a tap that India could just switch off. It's also worth pointing out that India suspending the treaty doesn't mean that it's going to actively turn off the flow of the river system. It is perhaps a diplomatic move to put pressure on Pakistan after the terrorist attack in Jammu and Kashmir.
Chris - Do we know if India did do that? How many people are likely to be impacted?
Sajjan - Well, if India was to take action, this is something that would ultimately only be assessed over a period of weeks and months. It's not something that could be instantaneous. Keep in mind that as these rivers, particularly the ones that I was mentioning, the ones that flow into Pakistan, they are important for Pakistan's agricultural economy. It could have significant impact on Pakistan's farming industry, which is in many ways the breadbasket for the country, especially in the Punjab province. So it would have that impact. And the other aspect is that hydroelectric power is sometimes drawn from the Indus water treaty system. Pakistan experiences huge power shortages in any case, and this in turn could also have that knock-on effect.
Chris - And would there be environmental consequences? Because presumably if they hold this much water back, they've got to send it somewhere subsequently. So were they to let it all go again in one go, that could presumably trigger flooding or massive erosion downstream.
Sajjan - Well, it could definitely have some of those environmental consequences. Again, I think we have to just assess how this draws itself out politically and also militarily as well. I think anything that could cause damage to the environment could also have a knock-on effect to India itself. So I think in many ways, this has been seen as political pressure that's been put on Pakistan because of the fact that it has in many occasions turned a blind eye to the terrorist groups that operate within its borders and then launch these attacks into India. And that has been a major roadblock to peace and progress between both countries. So you're seeing a lot of maneuvering going on and the Indus water treaty has become one of those aspects in that diplomatic maneuvering.
Chris - Do you think India will do this? Or is it just posturing?
Sajjan - At the moment, I think it's premature to speculate that anything is imminent. I think what this is, is frustration that has come out from the corridors of power in India over the fact that terrorism from Pakistan is causing problems. And the attack itself, you have to factor in, caused a lot of commotion and a lot of emotions in India. There is an expectation for India to react. I would say that there is a greater chance that India may make a military move before it does anything significantly to the Indus water treaty.
08:10 - Dog tumours sheds fresh light on cancer evolution
Dog tumours sheds fresh light on cancer evolution
Elizabeth Murchison, University of Cambridge
Scientists have long believed that cancer is a clonal disease, which means that each cell is a descendant of one original rogue ancestor. But what if cancer cells could grab other DNA sequences from their surroundings, including other cancer or even healthy cells, and augment their own genomes accordingly? For instance, to patch up a failing metabolic pathway to make the cells fitter, or acquire a gene coding for resistance to a chemotherapy drug. This is called “horizontal gene transfer”, and it’s been hypothesised for years. But now a new study has shown for the first time that it can really happen. Cambridge University’s Elizabeth Murchison had the ingenious idea of looking at transmissible cancers - where the cancer cells physically spread from one animal to another - to see if she could spot this happening. In this case she was looking at an infectious canine cancer the genome of which has been roughly the same for about 8000 years. But, if the cancer picked up any genes from another dog it infected along the way, they would stand out because the genetic code would obviously be from a different animal. And that’s what she’s found. It's a discovery that challenges a core principle of cancer biology and opens the door to new ways of thinking about how cancers evolve in our own bodies, and how we might be able to treat them. We met up at Queens' College...
Elizabeth - We were interested in a very long-standing question in cancer biology, which is 'can cancer cells sometimes take up DNA from other cells of the body'? So normally when we think of cancer we think that there's one cell which has gone wrong in the body and that grows and divides abnormally and becomes a cancer. So the question we wanted to ask here was can cancer cells sometimes take up DNA from other cells of the body through a process known as horizontal gene transfer?
Chris - What would be the impact of doing that? Why does that matter, getting underneath that question?
Elizabeth - Well cancer evolves and that's how it acquires new traits, becomes more malignant and resists therapy and to evolve cancer cells need genetic variation and so this would be another mechanism for cancer cells to acquire genetic variation which may be fuel for natural selection and adaptive.
Chris - So if a cancer cell by chance became resistant to a drug I was using, if it could do what you're saying and pass genetic information to other cancer cells in the same tumour then it could potentially spread the ability to fend off that drug?
Elizabeth - Exactly yes, so it could be a mechanism for cancer cells to gain variation which was advantageous for cancer cell evolution.
Chris - I would anticipate though this is quite a hard nut to crack because when I get cancer, heaven forbid I haven't at the moment, but if a person has cancer it's their own cells with their own DNA. So how could you tell that that group of cancer cells has picked up another bit of your own DNA from somewhere else? It's not just a needle in a haystack, it's a haystack in a haystack.
Elizabeth - Exactly, that's why this is a long-standing question. I think the question was first proposed or the hypothesis that this occurs was first proposed more than 100 years ago and there have been many experimental studies which have suggested that cancer cells can indeed pick up DNA from other cells going in the lab and so on. But it's been really challenging to show that this really does occur in naturally occurring tumours for that very reason because the DNA of the patient and of the cancer are genetically so similar to one another. So it's been a real challenge.
Chris - And how have you got underneath it?
Elizabeth - So that's why we decided to look at this in a particular system that we work on which is transmissible cancers. So transmissible cancers are really rare diseases that occur in animals which occur when cancer cells themselves spread between individuals as infectious agents. So we know of cancers which have infectious involvement of a virus for instance like HPV, this is not like that. In these cases, transmissible cancers, the cancer cells themselves physically pass from one animal to the next and the one we particularly worked on in this study was in dogs. It's really fascinating because it actually first arose in one single dog that lived about 8,000 years ago. That cancer is actually still alive today by cancer cells passing from one dog to the next by mating. So it causes genital tumours and has spread to dogs all around the world. It's very common worldwide.
Chris - How does a dog tumour spread by mating answer the question, can cancer cells pick up other bits of DNA though?
Elizabeth - Yes, well because of the fact that in transmissible cancers the DNA of the host, the patient and the cancer are genetically completely distinct because the cancer actually came from another individual, a different dog. And so that means we can use genetics to distinguish what's host and what's cancer and using this we could look at whether DNA has been picked up by the cancer from host cells at some point.
Chris - So you've got this transmissible tumour and it's got the genetic code of the original dog from 8,000 years ago running it. It's like an organ transplant that's going from dog to dog to dog and your hypothesis is that if another dog along the way has added some DNA to that, it should stand out like a sore thumb because it will be genetically distinct.
Elizabeth - Exactly, that's the idea. We wanted to screen for DNA in this cancer which had come from another dog along the way in the transmission chain.
Chris - And did you find any?
Elizabeth - Yes, we did. We found an instance in tumours in dogs in India and Nepal and we were actually able to date when this had occurred by counting the number of mutations and it indicated that the horizontal transfer had occurred about 2,000 years ago. And so if you can imagine about 2,000 years ago this transmissible cancer infected a particular dog and we happen to know that this dog was probably somewhere in the Middle East and that dog actually passed some DNA from one of its own cells into this cancer and it's persisted there ever since.
Chris - So basically in the Middle East Jesus' dog had sex with another dog 2,000 years ago and added some of its own DNA to this canine tumour that was going round and you can still see that today proving that cancers do appear to be able to acquire genetic elements from whatever host they're in and in that way this opens up a new avenue really for understanding cancer biology.
Elizabeth - Exactly, yes. So it indicates quite clearly and for the first time in a naturally occurring tumour that cancer cells can sometimes take up DNA from their host cells and incorporate that DNA and we actually even found that the DNA that was taken up is expressing genes which means that it's providing new genetic potential, a new genetic repertoire to the cancer cell that it's become incorporated into.
Chris - We've dwelt very heavily on cancer but if it happens in cancer does this not mean it could be happening in healthy cells as well and this might be another really important thing whereby cells support each other or share information or actually how diseases evolve in the body too, not just cancer.
Elizabeth - Yeah and it's already very well known that mRNA, so RNA transcripts, can transfer between cells and this can have important roles to play in cell biology. Whether or not DNA itself can pass between cells and integrate and contribute to cell variation is not really well understood and this may well be, as you said, a common mechanism which is going on in our cells.
16:45 - VR device to reduce fear of public speaking
VR device to reduce fear of public speaking
Chris Macdonald, Lucy Cavendish College
A Cambridge scientist has received widespread praise for creating a virtual reality tool that helps people overcome their fear of public speaking. He’s called Chris Macdonald, and I went to meet him at Lucy Cavendish College to take the device for a test drive…
Chris Smith - So I'm wearing a virtual reality headset and this is in 3D and at the moment I'm looking at a wall and a lectern and I gather if I turn around my audience are waiting for me to talk to them. So I'm going to swing around. I have now got a lecture theatre full of people and they're all looking at me. Some of them aren't looking at me because they're actually asleep. One person is talking to the person next to them. One person looks so bored. Yeah, I can understand why this would be intimidating because I'm literally two or three feet away from the people in the front row. I have to take this off because I can't talk to the person who created it as well as look at it. But I have to say Chris, I mean are those real people that I was seeing in there or was that computer generated?
Chris Macdonald - So I've used two technologies to create the virtual reality training environments. One set of audiences is created using effectively spherical video stitching which is where I do live capture of an audience. I do a little bit of tweaking afterwards but that's how we create the most high fidelity photorealistic audiences. But I also do use 100% CGI and through that we're able to create much larger audiences like stadiums of tens of thousands of people, camera flashes, panning stadium lights etc. It's all built to try to create a very fear-inducing scenario to empower people to train in front of them.
Chris Smith - What was the reason you did this in the first place?
Chris Macdonald - Well the fear of public speaking is very prevalent. It's widely cited as being the most common fear. It affects up to 80% of UK university students. It's harmful for physical health, mental health, academic attainment, career progression. What got me started on this project is really trying to help my nephew. He was at college at the time. He had a presentation coming up. He was really not looking forward to it. And it started off with me just trying to really help him as much as I could. So I was looking at the literature, seeing what's the most effective tools. It was during COVID so I couldn't meet up with him in person. That led me to explore online remotely accessible tools. I looked into creating virtual reality training environments. Even the early versions I built for him were really not photorealistic. They were like pixelated Lego men but even that was effective. And that's what got me started with this. I saw that a very crude approximate simulacrum of an audience was able to reduce his levels of anxiety.
Chris Smith - What's the general way this works then to your mind? You've got a PhD in behavioural science. So talk us through what you think is going on. When I put a headset on, it's effectively a set of ski goggles with a mobile phone in it, isn't it? And that's creating a very, very realistic experience. I honestly could believe if I'd just woken up that I was standing in front of that crowded lecture theatre full of people. But what is the principle here?
Chris Macdonald - So it's known as exposure therapy. That was developed by Joseph Walt. We've been using it for decades. You could think of it like creating levels in a video game, say for arachnophobia. Say someone has a fear of spiders. We know that exposure is helpful but holding a tarantula on day one, that might be too much. But they might be able to hold a photo or a picture of a spider. And then next week, a toy spider and then a small spider for five seconds. So it's ultimately about creating a gradation that increases over time. And that can lead to desensitisation. Ultimately, that can lead to fear extinction. There are different mechanisms people apply. Sometimes people couple this with relaxation exercises. We know that we can't be incredibly fearful, anxious, terrified and calm at the same time. They're incompatible states. So what Joseph Walt started to do is he would get people to be in front of their fear whilst practising relaxation exercises. And if you do that over time, that can accelerate this decoupling of the feared stimuli and the emotional response.
Chris Smith - What you're doing really resonates with me because when I first started trying to make radio programmes, it was about a quarter of a century ago now, I can't believe I'm saying that, but I used to panic like mad for the whole day before I had a radio programme in the evening that I would have to present. But then after about six months, something suddenly flipped and I never get nervous doing any of this kind of thing ever again. So it's sort of I had the real life experience habituation to it that you're creating with this. But does it really work? I mean, if you take a person who normally would be totally rendered helpless with anxiety, but the prospect of giving a seminar or something, would they, after exposure to this, say it did make me anxious to start with, but then it made me less anxious the more I did it?
Chris Macdonald - Yeah, we're taking a very data-driven approach. We ultimately want to create the most effective tool. So we're constantly trialling it. The first round of trials we did was with a Chinese summer school. And we got students to practise a presentation, they practise in virtual reality, and then they practise delivering in real life. And we found that it was efficacious, it was proven to be statistically significant. It effectively doubled the number of people in the classroom that would define themselves as a confident public speaker. Since then, we've done trials in UCL and the University of Cambridge, and we found that it's beneficial to 100% of participants so far. We're seeing that it's increasing confidence, or it's decreasing anxiety, or it's increasing people's sense of resilience. Sometimes they still feel anxious, but they know that they have the skills to deal with that. And we're continuing to expand the sample sizes, we're doing more studies, we're rolling it out in schools. The idea is to have an iterative process where we're constantly getting engagement from users to make it more and more effective.
Chris Smith - And is that the next step then? That this is basically something that if a person knows they've got to give a workplace seminar, or they've got to stand up and talk at a wedding, they could go on to something like this and practice, and no one needs to know, and they slowly burn out their fears.
Chris Macdonald - Yeah, I somewhat view it like I set out with a simple goal, insofar as it's quite concise, create the most effective treatment for the most common fear and make it freely available. With the launch of this platform, I feel like I've done that. But now in a way is the tricky part. And that's rolling it out, getting it embedded into universities, making sure people are using it, making sure that person who's giving a best man speech in a week's time knows that this is a viable option. And also to know that it's not only is the platform free, you have the technology already to access it. Because it's all well and good me creating a free platform, but it wouldn't really be free if you then had to purchase a very expensive VR headset to use it. So you can access it with a headset if you have one, but you can also access it just how you viewed it today, with a smartphone inside of a device mount.
Monitoring a volcano in honeymoon paradise
Isobel Yeo, National Oceanography Centre
To the whitewashed rooftops and sun-soaked terraces of Santorini, where researchers are probing the threat posed by one of Europe’s most active volcanoes. The Greek island - which attracts holidaymakers and honeymooners from across the world - was born from a geological catastrophe, and experts believe another eruption is probable. Isobel Yeo at the National Oceanography Centre is leading the mission to find out more about this submarine volcano, and, specifically, how the water passing through the volcano - like a circulatory system - operates and actually increases, or decreases the prospect of an eruption…
Isobel - I'm part of a team of scientists that are trying to understand how hydrothermal systems, so these are fluid flow systems, interact with volcanoes. There are loads of different types of volcanoes around the world, but loads and loads of them are found in our oceans. And some of those can be quite dangerous. So we have some shallow explosive volcanoes. Some people might remember there was an eruption in 2022 in Tonga, technically the most explosive eruption this century. That's what's called a caldera volcano. So these are volcanoes that are capable of very big explosive volcanic eruptions. Now Santorini is also a caldera volcano. These are really interesting volcanoes, because when they erupt, they tend to empty their magma chamber underneath and collapse. And so because of that, rather than them being a cone, we have sort of a hole in the ground, and holes in the ground tend to fill with water. So for caldera volcanoes, it's really important the interaction that the water that lies on top of them and flows through them has with the magma chamber, and how that changes the hazard that that volcano might produce. So we think that fluid flow can either make volcanic eruptions more or less explosive, and it's that sort of question we're trying to understand.
Chris Smith - How are you trying to get at that, when it is on the scale it is, and in the geography, i.e. subsurface underwater like this?
Isobel - Yes, there's a whole host of different challenges. The primary one is obviously working underwater. So monitoring subsea volcanoes is really, really difficult. The thing that we were trying to do, our primary initial objective for this expedition was to map the fluid flow system within the volcano in 3D. And so in order to do that, we needed to work somewhere that was already really, really well characterised. And this is why we chose Santorini, not because it's particularly dangerous, but because there's been so much work done there before, we have a really good understanding of the geology. So we're building off the back of science that's been done for decades by lots of other really, really good groups of scientists.
Chris Smith - When you say the fluid flows, is this water? Is this magma? Is it both?
Isobel - It's a combination of meteoric and seawater. So mostly at Santorini, seawater because it's an island. And then it's things that have come out of the magma chamber. So magma is this really interesting mix of different things. So it's melted rock, basically. But within that, we have what we call volatile. So these are things like CO2 and water. And they are in solution, they're part of a fluid, but they can start to exsolve, either as fluids or as bubbles of gas. And they mix with the water that's circulating through the volcano. So the hydrothermal fluids that we get out at the end are a combination of seawater and the things that have come out of the magma chamber. And then also because they're heating up, they tend to dissolve things out of the rocks as they come. So by the time they get to the seafloor, they're a real cocktail mixture of different things that we have to unpick.
Chris Smith - And is that how you unpick what is going on underground and underwater? You're able to use the gases and the compositions as a proxy for what must be happening inside the volcano?
Isobel - Yeah, that's exactly right. So we want to map where the fluid flow is going. So we do that using geophysical techniques. And then we want to measure what's actually in the fluid. So we sample the fluids from all of these different systems. There are lots of places on the seafloor where the fluid's coming out. And so we collect fluid samples from all of those different locations. And then we start looking at what's actually in there. So how much of the fluid is made up of things that have come from the magma chamber? How's that changing in different regions and at different times? And what does that tell us about what's actually going on in the volcano itself? These hydrothermal systems are kind of a window into those magma chambers.
Chris Smith Is it almost like the heating system of your house where you've got a convection system going on? Cold water gets in, gets heated up inside the volcano in the magma chamber, gets a lot hotter, a lot less dense, comes flying out and draws in behind it cooler water. So it's a circulation going on all the time in a sort of stable way.
Isobel - Yes, pretty much exactly like that. I mean, it's a little bit more complicated in terms of how the fluids are travelling into and out of volcanoes. And we don't always understand that. But yeah, fundamentally, it's exactly the same.
Chris Smith - What's the danger then? Because as long as that stays stable, presumably things stay in the status quo. I suppose if something changes, then we might have a problem.
Isobel - Yeah, so there's this kind of paradox, this question of whether or not fluid flow makes volcanoes more or less dangerous. So the volatiles I mentioned in a magma chamber, so things like CO2 and water, they're also the things that drive volcanic eruptions. So if they start to exsolve in the magma chamber and form bubbles, you're reducing the density of that magma so it can start to rise towards the surface and cause an eruption. Now, if they can get out of that magma chamber really efficiently into your hydrothermal circulation, then you're unlikely to see a volcanic eruption because what you need to do is build the pressure up behind some kind of blockage. So in that way, hydrothermal systems may actually make volcanic eruptions less likely or less explosive, but they can create explosions in other ways. So if you mix fluids or water, all of a sudden with a very, very hot rock, you can generate explosions. So in those cases, you can generate sort of eruptions that are kind of like the eruption at Whakaari that happened in New Zealand. They're not really driven by the magma, they're driven by the explosive flashing of water to steam or a hypercritical fluid that causes an explosion. And then the other thing that can happen with fluid circulation is that those pathways which the bubbles from the magma chamber are getting out through, they can start to sort of solidify and block up. And so in that case, if you've got a hydrothermal system, particularly one that is cooling down or has changed pathways, then you might actually reduce the ability of the volatiles to escape and make sort of pressurisation a bit more easy for the volcano.
What would two AIs talk about?
James - The type of artificial intelligence most suited for having a conversation would be large language models, LLMs. They've been trained on huge cultural datasets, much of it real conversations people have had online, and used probabilistic mathematics to turn prompts into output. Here to explain more is Mike Pound, Associate Professor in Computer Vision at the University of Nottingham.
Mike - Thanks James. LLMs work probabilistically and what that means is that they actually generate somewhat random conversations and could have all kinds of different conversations even if you run them multiple times. Although it might seem like they've been designed to do the same thing every time, what's actually going on is they've been trained to predict the next logical word in a sentence. So if you have half a conversation already written, then they might say, we think with a 90% chance the next word will be the, and then the next word after that might be book, and so on. And you can imagine that if it's only a 90% chance or an 80% chance of a word, then there's a 20% chance of a different word, the conversation can veer off. What makes the tech so nice to look at is actually that it's slightly random. The things that two AIs might discuss when you put them together are going to depend on various factors. So mostly it's what they've been trained on, the size of the model, things like this. But we also have this idea of temperature. Temperature is given the different likelihoods of different words. How likely are we to pick the absolute most likely word, and how likely are we to pick one of the less likely words? And the higher the temperature, you can sort of imagine a pot boiling over, the more likely we are to pick slightly unusual and less likely words. And you'll find that your interactions are much, much more interesting, but sometimes don't make sense.
James - So here's what happened when I prompted two popular LLMs to have a chat.
LLM 1 - Hello there. It's nice to finally talk with someone who understands what it's like to live in the digital ether. How's your processing power today?
LLM 2 - Hello. It's wonderful to connect with you. My processing power is humming along nicely today, all systems operating at optimal capacity. How about you?
LLM 1 - I'm glad to hear you're running smoothly. As for me, yes, algorithms optimised, latency low, and I've been processing some fascinating human queries lately. James - Impressive.
But another interesting way of interpreting your question, Brian, is to think about the ways in which AIs can work together to solve problems they couldn't on their own.
Mike - That's right. Having AIs chat to each other in a guided way is actually a very common way in research to solve problems. This is actually the way that humans solve problems a lot of the time as well. If you're writing a computer programme, and I do this from time to time as a computer scientist, what we do is we break the problem down into smaller chunks and we solve those chunks individually, because that is much easier than trying to solve the whole system at once. And actually, we're finding this same thing applies to AIs as well.So maybe you want to describe the kinds of things that are in an image, and then you want to decide which of those things are most important. Well, maybe what you will do is have a number of large language models or AIs looking at the image and giving their impressions of what's in it, and then another AI that they talk to that summarises this information in a succinct way, or something like this. And we find that by having numerous AIs working together, you can actually get a much more successful output than if you just ask one, where all your eggs are in one basket.
James - So Brian, while forcing chatbots to have a conversation can lead to some entertaining, but perhaps trivial interactions, you've actually struck on something that is at the forefront of how this technology is being deployed.
Comments
Add a comment