Technology, Connected - How AI Satellites Detect Wildfires and Monitor Earth in Real Time

Episode Date: February 20, 2025

Fintan Buckley, CEO of Ubotica, joins Thinking on Paper to explain how artificial intelligence is changing the way satellites observe Earth, detect disasters and make decisions in orbit.Traditional Earth-observation satellites often collect data and send it back to ground stations for processing. Ubotica is developing onboard AI systems that can analyse images in space, identify relevant events and transmit useful information without waiting for the full dataset to reach Earth.This could reduce the time needed to detect wildfires, monitor environmental changes and respond to maritime incidents. It also raises questions about surveillance, military use and how much decision-making should be delegated to autonomous systems.In this episode, we discuss:How AI-powered satellites workHow satellites can detect wildfires in near real timeWhy processing data in orbit can reduce disaster-response timesHow satellite AI supports maritime security and environmental monitoringThe privacy implications of increasingly precise Earth observationHow autonomous satellites could be used in military operationsWhether humans should remain involved in critical decisionsWhy removing old satellites won’t solve the wider space-debris problemHow realistic fictional surveillance satellites areFintan explains where onboard satellite computing is already being used, what technical limitations remain and how close the industry is to deploying more autonomous systems in orbit.The conversation examines both sides of AI-powered Earth observation: its potential to improve disaster response and environmental monitoring, and the risks created when satellites can analyse, prioritise and act on information with limited human involvement.Please enjoy the show.--📌 Stay Connected:🌍 Website: ⁠ThinkingOnPaper.xyz⁠ 📩 Email: ⁠[e-mail]⁠ 🎧 Podcast: Available on all platforms🔔 Subscribe for sharp, no-fluff insights on AI, space, and the future of technology.

Disruptors and curious minds, welcome to another episode of Thinking on Paper, where we take you, the curious, entrepreneur, thinker, worker, doer, behind the scenes of the latest technology that we're seeing, where we see the world going, and peek behind the scenes a little bit, figure out what it means. What are we getting into today? Today, Jeremy, we're speaking about a technology which, and no understatement, might change humanity, save civilization. As you know, people, we're reading Nexus in our book club. Check out thinking on paper.xyZ for the second pillar in our framework for understanding emerging technologies. And listener, it's been a story of doom and gloom in Nexus of late. we've been talking about the nefarious dark side of AI. And today I'm very happy to say the positivity is back, the optimism is back.

We're talking about AI in a good light. We're talking about faster disaster response times, environmental protection, maritime safety, precision agriculture, national security, urban planning. I don't know that the list of the possible benefits of the technology we're speaking about today, almost knows no end. Of course, there are some disadvantages. We're going to talk about that. But on the whole, it's an awesome technology, Jeremy.

What are we on? We're going to space. Space, space. So real quick, before we introduce our guest, I've been a fan and a nerd of Carl Sagan's perspective of the pale blue dot, which is basically looking back at the Earth from way far in space and seeing this giant thing, this giant rock, that we're sitting on as a tiny blue dot.

So I'm excited to learn from our guests. What is the most important thing we can learn from a space perspective on things? So without further ado, let's get into it. And I'm asking questions and our guest doesn't even hear yet. Mark, how about an intro? Yeah, bring on Finton, Buckley from Ubotica, AI, satellites combined. We're going to find out more. Welcome to the show.

Fenton, thank you for thinking on paper with us. Great. Thank you, Mark and Jeremy. Thank you for having me on the show today and looking forward to the conversation. Yeah, so we sprinkled Carl Sagan into the conversation already. Let's start with that first question with like, what can we learn from the perspective of outer space about the planet Earth, about the systems, about the climate, about all of that stuff? What's important from the view from space for us? the most important thing that we have from space is through various modalities we have the ability to have 24-7 365 days of the year constant surveillance of the earth and from that develop the technologies to provide the systems, the structures to keep our earth safe.

We are going through a time of tremendous change in terms of our climate, in terms of, you know, geopolitical issues that are going on in the world. The world is changing dynamically. And we have the ability through satellites, and particularly Earth observation satellites, to observe and to take action in near real time to the events that are happening around us. You know, you think about, you know, what are the pressing issues in the world today?

Climate is a big one. You know, you look at, you know, the fires that we have in Los Angeles, which are actually very personal to me because, A, I lived in Los Angeles. My sister lives in Los Angeles. She lost her house in the fire in the palisades. with Earth observation technology, you know, we can see and observe in real time exactly what's happening in an event like that to assist in to supporting what's going on on the ground in terms of fighting the threat that's happening in real time. We have a lot of issues around maritime security. What's also very close and personal to us here in Ireland is we are the gateway for a lot of the telehealth.

color communication infrastructure that comes across the Atlantic, right? And we have to be able to defend that, right, and protect it against accidental and nefarious harm to those cables. We can see these things from space, and we need to be able to react to these threats in real time. And that's really what your body is all about, being able to deploy technology onboard satellite to react to these events

and to provide systems and structures to those who are going to respond to the humans who are going to respond to this. Let's double click, as they say, on the LA wildfires and I'm sorry to hear about your sister losing her house. I hope

that it's the healing is coming along. Okay. If we double click on those fires, And from Eubotica, what are Eubotica doing? And if, say, five years in the future, what would you have done or could you have done

to make those fires less impactful, less dangerous? What could you have done to have helped more? So from satellite, right, with the persistent coverage, you can detect and predict the movement. than the fires, right? And that early warning systems you can provide to the fire responders. That's the key thing that you can do

because you've visibility, right, from space to actually provide that level of warning. Now, granted, you know, what happened in the palisides was very, very quick, right? It was, you know, with the winds and everything, it moves so fast. But with the ability to see this stuff happening in real time, you can help in turn to the evacuation

and the deployment of the resources on the ground. And a key part of that is with the satellite, it's not just doing the imaging on board, but actually doing the processing on board as well. So it's not getting images down. It's actually getting the insights for the direction that the fire is moving, the density of the fire, all of the such of the stuff. That's the type of thing that you can do

when you employ AI technology on board these EO satellites. That's a really important thing to think about too because, you know, getting images and then sending the images down to Earth and then having those images or that kind of data processed on Earth, that's time lag. That takes a lot of time to happen. But if you're real time processing all of this stuff, let's think about this like turning data into information, right? We humans don't use data very well. We use information very well, right? So having it, so how much does that speed things up? Like how?

Did I read correctly that the cloud cover actually slows down a lot of this? Is that true as well? It's about multi-modality, right? Because yes, cloud cover is an issue. You, Mark, in France, it said that France would have much less cloud cover than we would have in Ireland. So it's multi-modality. So it's using, you know, optical sensors. using IR sensors, and it's using SAR sensors as well.

So synthetic aperture radar. So it's a combination of technologies that you need to be able to generate this, this persistent observation, right? So, you know, it's complicated. It's not cheap. But the benefits of doing this are well worth the investment in doing this. And then when you apply the onboard processing on top of that, you, you now get, you know, real-time responses to what's going on.

And Jeremy, you asked about the, the latency from image to inside. So, you know, one of the key things that we have done is we have invested in our own satellite to prove out this capability. And we have demonstrated that we can go from image to insight to ground within five minutes. You use the AI to process the image

and then use inter-satellite links to get the output of that information down to ground. So that's the key. You're dealing with five gig images. These things are big, right? And then you've got a way to get over a ground station to actually get the information down. And then you've got together from the ground station

to wherever the processing economy. So it could be hours to days, right? So, and when you're dealing in a disaster scenario, you can't afford to wait. You need to get this information. Well, in the case of the wildfires, too, like things, you mentioned the world is changing. The environment changes so rapidly. So by the time that old image gets eventually processed, we're onto something new already. So having that real-time access is really important.

Yeah, and then another example of this is, you know, in a similar type of scenario, you know, you can use satellite. satellite imagery, process it on board the satellite to figure out, you know, what's the best route to get out of a situation? Maybe not so much with fires, but, you know, in flooding and things like that, like in Valencia, you know, where's your road network after this event? Imaging on the satellite, you can actually process the image and determine, you know, what has changed in your road network, right, before and after the event and to get that down to the responders. So, you know, the applications here are really significant and the benefits are huge. Yeah, so we talked about the real-time processing and kind of this idea of, you know, edge compute, edge AI to process these information, these sets of data turning them into information. I ran across the term in our research, live Earth intelligence, which I think is amazing. That's really, really cool.

It's like we're tapping into the subconscious of the planet in a way. It's really interesting. I want to talk about dynamic targeting. Because a lot of things happen quickly and they change quickly. And if you're looking at one spot or if a satellite's looking at one spot, how do you quickly make sure that moves or moves based on a tracking pattern? So you have some partnerships that you've put together, right, to make that happen? Yeah, just a segue into that.

So, you know, the name of the company is Eubartica. And where that comes from is ubiquitous robots, right? That's where we made the name from. And the end goal is to have these satellites, to have them autonomous, right? To turn them from today, they're essentially dumb cameras, right? They will image, you know, based on input that's coming from the ground, where, you know, a human, you know, does the orchestration and the station that says you will, you satellite will take an image when you're,

over this part of the Earth. So part of the vision of live Earth intelligence is that the satellites make these decisions themselves. Okay. And what we're working with, sorry, distracted. If we're working with JPL there in Pasadena and the US on this concept of dynamic targeting. And this encapsulates a number of concepts.

The first concept is, where your operator on the Earth schedules the satellite to take an image over a certain part of the world. Okay, certain area of interest. And if that area of interest happens to be cloudy, you will end up taking an image, consuming valuable satellite resources, primarily the power budget because you're constrained and how many images you can take during an orbit. okay, so you will waste your power budget on an image that's useless. So the idea of dynamic targeting is, first of all, to have a look-ahead capability using the same image sensor to look ahead and to take a low-resolution image and say, and then use the AI to process that and say, is the area of interest coming up cloudy or not? Okay.

And if it is, then you reorientate the satellite, okay, with no human intelligence. intervention and oriented to send the image or on the satellite to take the the the image of the area that you're interested in. So that's the first part of the dynamic targeting. And then we segue that into, you know, which we call this the Nasta, the novel observing strategies is that once you have taken an image of an area of interest, you analyze that image. And now you know it's something of interest. Okay, so this is a trigger. And now you want to marshal other satellites in your constellation, right, to all start looking at this area of interest.

So using either satellite inter-satellite links or using ground stations, start to bring other resources in the constellation to focus on the area of interest. So you imagine, you know, you detect, you know, in a maritime situation, you detect a vessel that is an effect. Right. So now you want to keep all your resources in your constellation, monitoring that vessel as it moves. So that's kind of the second part of the dynamic targeting is actually using other other satellites resources to maintain constant vigilance on the area of interest. So the so the nefarious vessels, I got a I got a kick out of this when I was doing my do my show prep the idea of dark ships, right?

So I don't know if you guys have teenage kids yet, but you know, there's this thing called life. 360 that's on phone so where you could see where your kids are and that if it happened in my day it would have been such a pain in the butt but they disable life 360 and they turn into dark ships right and you don't know where they are and you have to call them and but i thought that was that's just a silly analogy but in in my house dark ships are called pirate ships could i just could just back up a little bit and play my my stupid card just i want to be clear on of what Ubartica is actually building. Are you building the satellites?

Are you building the software that goes into satellites? Are you doing both? Can that software be used in Starlink satellites or the Amazon Kuiper satellites can that be used everywhere? What's the actual technological stack here? Yeah, so when we start off on this journey, you know, the key brains behind the, the company were involved and founded the first company that actually developed what's called

a vision processing unit. So this was a piece of silicon that integrated the ability to take images plus the ability to run AI applications on those images in the same piece of silicon. So the very first, what's called a BPU, right? The company was acquired by Intel in 2016. And at the same time, there was an opportunity perceived to take that same silicon technology and actually look to deploy it on board satellite. That in itself was fairly novel because, you know, historically, when you're developing silicon systems for space applications, you design them to be resistant and resilient to the conditions that you expect to experience.

in space in terms of radiation and what have you. So we work with the European Space Agency to qualify what's known as Coth Silicon, so commercial off-the-shelf silicon, to qualify to assess how it would perform in space before it actually got flown. And then through the qualification, we assured ourselves that it would fly and it would perform with very little mitigation. And so we were the first company to do that. And so our initial focus was developing the hardware and the software

that will allow us to run AI algorithms onboard satellites. Fenton, I just want to clarify one thing. So you and this group that you were with were the first people to test and make sure commercial off the shelf silicon doesn't fall apart and can actually function in space. AI processors, yes. That's big. Okay, cool. Continue. Absolutely.

And now, since we have done that, there are many other people doing it and saying, hey, the guy took the pain and now we're following on. And so now what we, you know, what we're much more focused on now is not just the deployment of the platform to run AI applications, but actually is developing the products, the end user solutions for, for our customers. So our customers, you know, with naval services and things like that for maritime security. So that's where our focus is going now. So is that moving towards like, all right, so we have the ability to have AI processing on satellites in space.

And now it's like, okay, what are some of these use cases and how can you make the user experience easier, or the app or some kind of delivery of that service to these folks. Is that what I'm hearing? Yeah, that's exactly right. And I wouldn't call it an app. It's more a service, right? So it's the ability to provide information. And I think you said it yourself earlier on, Jeremy,

we don't want images, we don't want big reports. We actually want information. consume and we want us when we need it. And so the services that we're envisiting is banal to monitor particular areas of the earth on a continuous basis, knowing using the onboard AI when an event occurs, right? And then on ground mining, you know, all of the traditional databases of EO data and to provide a, you know, narrative, the history of what happened up to that point, and then using the satellites on orbit

to then provide the ongoing monitoring once that event has been triggered. That's the service that we are offering, that we are building to offer, I should say. Well, with the naval, with the national water security, I can envisage that surface. I can see the boats on the ocean. I can see the real-time information saying there is a dark ship here, there is an oil spill here, there is this here, you need to go here, here and here. I can visualize that and I can see that being in operation very quickly. When you were speaking earlier about the floods in Valencia, the wildfires in Los Angeles, especially the floods, and you were talking about kind of escape routes almost where the satellites can view what's happening in real-time. And I was

in my silly head imagining an almost like Mission Impossible scene where someone is on the streets in Valencia, the water's coming in, and they're like, the satellite says, go left, go left, you go left, and they go straight on up here, up those steps, right here, and describing that escape process almost. And that's a much further, a long way off compared to the service of, or is it not? It's not, Mark, and that's actually one of the easier applications to, to implement in AI. It's, you know, the, you know, it's basically you're doing a before and after comparison, right, on your road networks within the city. So that is actually a fairly straightforward application. As long as you have, you know, you need your assets over the area

at the time. That's where Jamie's friends' ways comes in. Maybe there's some kind of layer upon an app like ways, which makes it accessible for these more human everyday use cases. All right, that's brilliant. You know where you got me thinking in that regard? I don't know you remember the show with Todd Hazlehorse with Heel, H-E-A-L-E. I do remember logistics, the final mile.

They have a blockchain-based logistics company where they track things from source to final user, right? And like this middle mile thing, right? That would seem like it could be an interesting application of like tying in, you know,

some of these route optimizations or, you know, route optimizations based on weather, based on, you know, shipment and that sort of thing. Do you ever think about logistical cases as well? You know, what we're really thinking about is, is situations where it is hard to impossible to have ground-based monitoring.

So, you know, with vessels, you know, with logistics, with containers, you can put trackers into containers, right? And, you know, that's an easier solution. than actually using satellites for the observation. With us is more about it's the unexpected, right? It's the how do I monitor my maritime? How do I react very quickly when there's a climate issue of some description? How do we react to that and get visibility and be able to make decisions?

And also, it needs to, from a commercial's perspective, It needs to be, you know, an application, a requirement where real-time information is of key. You know, there are lots of use cases for satellite data in the agricultural space. But in those scenarios, in many cases, the real-time element is not important. So if you're getting your data a day later, it actually doesn't matter. And then you know, you apply your ground processing and make your decision. So it's really about... You're saving your power...

Yeah. Saving your power budget there, too, because you don't need it. And that's a detective rush. That's, yeah, yeah, that's correct. Let's go back to the wildfire piece, because I know that's it. I mean, it's just a problem that never goes away and it gets worse and it gets worse. And so I don't know a lot about, like, the, what people track related to the potential or likelihood of a fire happening. But I get an analogy in my head, like what we do today for tornadoes, right?

and there are predictive things that weather can use for tornadoes. Okay, we're starting to see a little cycle happen. Here's a warning. Here's a potential likelihood. Are there some of those data points that you could monitor in real time in high fire, high wildfire areas that you could create these early alerts for that? When we talk about the optical images that we would work with, They are, they're hyperspectral.

Okay, so you have multiple wave bands that you can, you can analyze, use to analyze the situation. I mean, typically we use the RGMB because they're the optical ones. They're the ones that we humans can see. So your question is, can you analyze weather conditions to predict events beforehand, possibly, right? But I would suspect that the model. that's done by meteorological services on ground is probably much more efficient at doing this. You know, going back to, you know, the wildfires of Pasadena wildfires, and forgive me my harping on this,

but, you know, I know the area, but, you know, I was sitting at home on my couch and my sister, WhatsApp me, she said, just to let you know, I'm leaving the house, there's an evacuation order. The people on the ground, she had no idea how serious the situation. situation was, right? But if you observe it from a satellite, you could see, right, in time shift, how quickly this fire was moving and how dangerous the situation was. And that's where satellite imagery, EO data and applying AI onboard the satellite to get the insight down to ground as so quickly as possible. That's where the real benefit is, right? And even, you know, another minor case, when people were evacuating from the palisades, because,

because, you know, it's a, it's not your typical area of Los Angeles. It's not freeways going through the middle of it. It's, you know, it's small streets. And traffic was getting backlogged and stuck because you had so much traffic, so many cars trying to get out of the area at the same time. A satellite overlooking that, right, could actually help to, to root the traffic out of the area rather than people are getting stuck and gridlock or whatever. So these are all of these things that you can do if you have the satellites on orbit looking at the scene.

On that, maybe we talk about connecting the dots and holistic approaches to technology here and how technologies can work together to solve a problem. Perhaps AI on its own isn't specifically for wildfires. The solution, rather than that top down, you have that bottom up where there's some kind of infrastructure on the ground where people if we've spoken about this in quantum season about posting on social media maybe there's these patterns that emerge in the data of social media or the

texting or the WhatsApp or information about you know sort of changes in the pressure changes in the wind changes in certain aspects of nature which kind of then a quantum computer picks up on and feeds it to the AI and the satellites and the satellites come over and they do that imagery testing and then a bit like the dishes, then you have that last kind of mile problem where getting the information to the people on the ground who are in danger. Yeah, I mean, that's the the ISL link. So you're tapping into your Starlings and your, your, your, your Kuiper networks,

right, to get the information down through essentially the internet down to the end user as quickly as possible. I mean, you're not broadcasting this information, but you're getting it to the first responders, right? And then, then they decide what they're doing with this information. You know, you don't want this going to everyone's phone saying turn left at the street, you know, at the end of the street. You don't want to instill panic in the ground population, yeah. Exactly. That's exactly right.

Yeah, Fenton, that's been a theme on this show for since Mark and I started, since episode one, just this balance between hierarchical systems and emergent systems, top-down, bottom-up and like where the successful handoffs come into this. And I think solving that or at least trying to figure out how that is incorporated is probably going to be a big thing in what you do and how you apply some of this stuff. A lot of what we're doing is trying to take the human out of the loop, particularly with the dynamic tasking and the novel of observing strategies. But at the end of the day, we always do need the human in the loop. You know, and I know we're in this era of mass deployment of AI, questions over AI. I personally think we always need the human in the loop somewhere.

Yeah, we're saving the human. The humans at the end being saved. That's the part they pay is the unwilling or willing victim or savior. Again, a massive, massive theme, Fent, we actually broke down, I'm not sure if you've ever read Pachy McCormick's not boring newsletter. We broke down his call to humanity related to AI and how you can prepare yourself for the AI revolution. It was a great breakdown that we got involved with.

And then we talk about your thoughts are completely aligned with our, we call it our host and author of our book club. Noah, Yuval Harari, as we talk about sapiens and this idea of keeping humans in the loop. But Mark, I think it's time for hot buttons. Yeah, let's do the hot buttons. So this is five questions, 30 seconds, system one thinking. There's no wrong answer. We just have to finish the five.

questions in 30 seconds. Have you got that clock? Can I see the clock? Star Wars or Star Trek? Wars. Is Elon going to Mars, yes or no? No. Crop circles, aliens or farmers? Farmers.

What's the best book on technology, fiction or nonfiction? Sold a new machine. When you look up at the sky, what one word describes how you feel? Wonder. Let's talk about wonder. are deficient in our sense of wonder whenever a question is asked we go right to our phone right to our computer and get the answer sure like that makes sense if we want to know you know how quickly

how long a drive it's going to take from Atlanta to Charlotte what time the next movie is but man like how can we enable wonder in people like what what do we do with that I know it's a bigger question but how would you enable wonder in anyone you're talking to I think my generation is responsible for this. I think our technology, you know, which we have developed over the past 40 years, has actually caused us not to think,

not to have quiet time, not to have downtime. And I think that's a problem. Do nothing. Sit on a chair and just pond and think. And I think that's the problem. I'm guilty of myself.

I mean, if I have two seconds, you know, take out the phone and look. And I think it's a problem. You know, social media, which I shy away from, I think is a real problem because it just occupies the mind. And sometimes we just need to sit and think. So how do you inspire that? I don't know. Take away the phones.

Shut off the internet for an hour every night. Maybe something like that. Don't be bored. Whatever you do. Don't be bored. Don't be bored. Well, great. Thanks for entertaining that question.

I've just been thinking about it a lot in a sense of wonder and that sort of thing. All right. So next, we are on to thinking on paper news. A couple of things that have been in the news lately that we've been talking about a little bit ourselves. We'd love to get your thoughts on. The first thing I've been tracking for quite a while, but there's a lot of news out there about the potential. potentially crowded nature of lower Earth orbit.

And, you know, that there are even technologies out there that people are designing to help remove space junk. Like, how big of a concern is the congestion of lower Earth orbit at the moment based on what you know? Got some numbers for you there, which might not be in date. Starlink, November 24, had 6,676 satellites up there. The Space Development Agency 200 Amazon Kuiper Systems 3. thousand. You've got the small satellites market, India, China. There's a lot up there. Yeah. I think what we need to think about, particularly in the low Earth orbit, is that all of these assets have a finite life cycle, right? Three to five years. So I personally don't think the problem

is going to be the number of operational assets, but it's the, what happens, those assets when they are no longer operational. And, you know, that is a big threat. So how are they are deorbited or how are they retrieved from from the low orbit?

And, you know, there are companies that are, and you know, this is, you ask me, you know, Star Wars or Star Trek, one of the things that really maybe stopped back and say, wow, this is an amazing time that we're in in the space industry. When you've companies talking about developing space tugs, refueling, reorbiting, you know, all of this technology. In your mind, you can see in 10, 20 years,

space is going to be just like a harbor, just going to be like, you know, San Pedro, in Los Angeles or whatever, you know, that you are going to be managing all of the assets that are in space. So I'm not overly concerned at this point, right? But I think it's what do you do when the assets are decommissioned is a bigger issue as far as from what I can see. So this, this again, I'm going to sprinkle up in a couple of things that I've read over time that are very pertinent to this discussion. If anyone's interested in space and how things are progressing in space. There was a book written a long time ago called The High Frontier

in the 70s that Gerard O'Neill, I think, or Gerald O'Neill. And it was a plan to basically build human colonies in lower Earth orbit, fascinating, approved by NASA. NASA was like, yeah, this sounds great, but no one ever funded it. It was possible, but no one ever funded it. But that's a great book. And I've got a free business idea brought to you by thinking on paper, you by the power of three disruptors and curious minds. So if we need to figure out how to deorbit some of this stuff, we need to build satellites that over three to five years start to just, start to just fizzle away. Just kind of like that spy message, hey, this message will self-destruct in 30 seconds.

This satellite will self-detruct over the next five years. Isn't the problem that even microparticles are a hazard? So you don't need to, you need it to really come down. in one piece. Can they not be just fueled? They run out of fuel or run out of energy and whatever and then just fall to earth and burn up? Well, I think the biggest problem is the batteries, right?

That you can't recharge them. I think that's the problem. But actually, I think it's not about how, you know, we destroy them. I think we have to be thinking about reuse, right? So it's servicing of these things. you know, I saw something recently about one of NASA's long-range missions that's gone out to outer space. I can't remember which one it was, but they talked about, you know, the redundancy and resilience they built into the satellite.

So they had failbacks and failbacks, right, because they couldn't predict what was going to happen, you know, in over the lifetime of the satellite. It turns out they didn't need the redundancy in the system because the, you know, the, you know, the systems and the design worked really well. So you have these satellites. They're designed to last, right? Why not to service them and keep them going again? And I think that's the way we should be going rather than thinking about the deorbiting and destruction of them.

Because the biggest costs and the biggest impact to Earth is actually getting the things up there, right? So once they're up there, let's use them. Just on that, how much is that coming down in price getting them up there? I think you'd want to ask Elon must. I think SpaceX have really got this down to a T. Do you know him? Can you put us in contact? No, I do not. I do not know him.

But, you know, it is relatively cheap and compared to where it's been. It is cheap to put them up. But it's, you know, it's the environmental cost to putting them up as well. Let's transition from the kind of say the hardware perspective back to the software perspective. So in talking about AI, another thing that's been in the news quite a bit, and Mark and I've been unpacking and talking about a lot, Google AI has changed their principles. And some of the things that have come out of their documents without a lot of notice have come out of their documents stating that, you know, use for surveillance, use for wartime stuff. use for things not great for humanity are now suddenly out of the principles. But as Mark highlights every time we talk about this, they still have the pretty butterflies on the document

and everything looks great. What do you think the implications of changing philosophies like that this early on in the game? It is a conundrum. And we have, you know, even though we're, you know, We're a small company. We have had to ask ourselves the same questions. You know, what if someone came to us and said, we want to use your technology in an offensive military scenario? And we have made the decision that that would not be something we would be willing to support. having said that, I think it is easy for us being a small company in a usual country,

which historically has never gone to war, has never been aggressive to take this stance. I think it's much harder for larger countries to take that stance. If you're operating, so for Alphabet, headquartered in the US, the priorities are different. And I could not criticize or condone decisions that are made by other companies in other countries. There are lots of factors of play. So that's a long wind of the way. It's a long wind of way of saying, I actually don't want to comment on this.

So. Jeremy, a question for tomorrow's book club where we are talking about democracy and AI, part nine of Nexus. And that's a very interesting point that how does the history of a country dictate what it will do going forward? And I think that's very interesting because the historic and geopolitical history of a country will influence their philosophy, their ethics, their principles on what happens next. So yeah, we'll talk about it more tomorrow. Absolutely. And Fenton, Fenton, I think, I think you, you answer that question actually really well and generated a new perspective in my mind on it because, you know, you go back to, you go back to like Cold War, right? U.S. and USSR, right? And everyone's like, oh, we're there building this. We got to build this and we got to build this and we got to build this. And there's a, there's a, there was a real kind of fear back in the day of that kind of stuff. So, you know, staying on top of, um, stay on top of certain things, you know, can kind of make a. bit of sense. It does scare me when things get yanked out of that, but there's always the other perspective that I don't see. I'm not the head. I'm not the chief guy making sure the US is safe

every day, right? So I appreciate that perspective. But also you know, the the opposite concern is also there that, you know, we're in this agentic world, right, where you have, you know, AI agents, right, generating code and generating other agents. So, you know, we are getting, very close to, it feels like, the point where machines are going to be doing things and we don't know why they're doing them. You know, we already have this problem with models but trusted AI. But now you layer on top of this, models making more models, right? You know, it's a concern. Yeah. That Russian doll of AI agents teaching other AI agents what to be, what they should be doing, what their incentives are, what they're aligned to do, yeah.

Like it's going to be really tough to find the audit trail on that. I think. And then will you believe the Argo Trail, right? Exactly right. Yeah. So again, we keep pointing back to this book that we're unpacking Nexus and it talks exactly about some of those things that are just really, really interesting in the financial system. Think about the financial system. Like, we don't, like a small percentage of the world knows how the financial system like really works. That's why we got into trouble in, you know, the financial crisis recently and that sort of thing. But what about when like agents upon agents are creating things financially that we can't even begin to understand. So that's like it, we could sit here and talk for hours about all this

stuff. It's super intelligence. They'll be building satellites. Can I just bring it back to satellites to with a soft, silly question? When you're watching James Bond and the satellites, they're moving the satellites from MI6 headquarters in London and they move the satellite and they zoom in on some guy who, who's carrying a map on a phone in Lisbon, and they go right down and they can read it in his hand. When you watch those kind of films, do you think, oh, that's really realistic, or do you think that's completely impossible,

and this is just make-believe? Like, how accurate are they in terms of what satellites can do today? If you'd asked me that question 10 years ago, I would have said, it's pure fiction. The capabilities of the satellites today, is impressive and getting more impressive. And there's two elements to this. Could they read that?

Could they read that from space? Just for our listeners who aren't watching, I'm holding a piece of paper with some words which are written in maybe font 12 or 13 on a computer for your reference. Maybe, right? Because you have two different, you know, sectors in the Earth observation.

You have the military sector and you have the commercial sector. you actually don't know what the military sector are capable of doing. And so when they say maybe, if not, it's pretty close. I mean, you know, the satellite that we were working on right now couldn't read that. You know, we would have a resolution of about five meters. So each pixel, it would be the equivalent to five meters. But we're getting, you know, we're getting to the one meter and less.

So it's coming in the commercial space. It may already be there in the military. point back, Finn, like, what you guys are looking at and what you guys are trying to do with live Earth intelligence. I mean, I think, I think that's really a compelling thing, right? Especially if we want to solve these big problems. We need to get the information. We need to get the information when it's usable, not when it's old, and apply it to things that we can make quick decisions about in real time. So, Finn, I think I want to re-re-hit on that. I think what you guys are leaning into is really, really cool. Mark, let's hit the carryover question, and then Fenton, if you could leave us a question,

and we'll try and land this satellite. Yeah, so the carryover question from our last guest is a nice, easy one for you. Fenton, it is how should we be thinking about data today so that it is still private in a thousand years time? I think we shouldn't be thinking at all, because I'm not sure it's going to be relevant at all. in a thousand years time. So don't worry about it. It's of no relevance. Why not? Because data won't exist.

We'll all be dead or another reason. I'm not sure it would be of any relevance and any interest to us. You think about how the world has changed in the last thousand years, right? Go back a thousand years. Is there any data that it would be relevant to us now, except for historical, right? It would be nice to know. But actually, do we need it? No.

It's not something I would do sleep over. So in a thousand years, Mark, maybe we move to an omnipresent super intelligence that we're all real-time connected to. And we don't need to store anything anywhere. It's just always everywhere. That's what we call a pregnant pause disruptors and curious minds. All right. So carry over question for you, Fenton, anything you're thinking about, it could be on any, like we talked about, anything from surfing to what pasta you prefer, to technology, to, you know, something. you read yesterday that it's puzzling you.

My head was thinking technology. I would like to ask the next contributor. What do they think is the next invention that's going to impact our lives as much as the mobile phone? For all our listeners and our next guest. Maybe nothing. Finn, thank you so much for joining. us today. I think we have some great insights as to as to what you're building. And we'd love you

to stay in touch with us and keep us posted as you're building these these solutions for these different industries using the technology that you've already proven. Stay in touch. It's been great chatting with you. It has. Thank you. And for all our listeners, stay tuned because we're moving into Star Trek season. We will be interviewing and speaking to more people building more interesting things in space. So stay tuned, thinking on paper. at X, Y, Z to learn about all of those. Sign up.

Friends and neighbors, we are now backstage. We just had a conversation with Fenton Buckley, with Ebotica, talking about real-time data processing, image processing from satellites to enable better decision-making, quicker decision-making, being able to tackle really tough problems that have a lot of data points. And we as humans struggle with things that have multiple interdependencies, right? That's what I think AI can do better than we can is pull all of these things together. That's how the computer that beat, you know, the best

go player in the world figured out something crazy outside of the pattern of what we think, right? So, Mark, first takeaways, first and what did you learn today that you didn't know about before we got here? I have to tell you about my holiday last year in order to get to that point because last year I was on holiday in the south of France watching shooting stars with my daughter and my brother and how about 10 o'clock half 10 at night the sky went a bit like the blue of my background there was a streak of blue indigo bright light that shot across the sky and for a moment. moment we thought they were here they've come finally it was very eerie very strange very unexpected um obviously we very quickly worked out that it was starlink deploying satellites into low earth orbit and i went on the internet and found out how many hundreds had just been deployed and that's what it was just passing overhead and that's ever since we saw those i've been

eager to have some a satellite expert on the show to tell us um honestly what did I learn? Beyond the technicalities of it, I learned to remember about the positive, optimistic side of what is possible with artificial intelligence, as I said during the show after eight months and eight episodes of the Nexus, to think about AI being used for disaster response. We've all seen the fires. That's helping people on the ground for when an oil spill might happen,

and getting people there because they know about it because they satellites and the A of work together to know it. Maritime safety, urban planning, like the positive things, the life-changing civilization, extending humanity, whatever that word is, the kind of hope of the technology. So that's what I learned. I learned to remember. That's great.

That's great. And I think about it, if you've ever written a song before. That's great. No, it is. I think it's great. So if you've ever written a song before, And, you know, I've, I've written many, many bad songs. And writing a sad song, writing a darker song is easier than writing a happy song.

And we really think about this. Like a lot of the magnetism for discussion points related to AI, it's easy to think about all the scary stuff. It's easy to think about like, oh, man, what if this stuff gets in the wrong hands? What happens? But it's really cool to highlight people thinking about this differently and applying the technology to change the world, right? To be able to enable this real-time data, right, and turn it into information and keep humans in the loop to use that information to make good decisions. That real-time Earth intelligence was something that is important. Maybe I knew that cloud cover is very detrimental to the flow of information between,

satellites on Earth, taking it, you know, if it's cloudy day, two days, but maybe with this real-time Earth intelligence, with AI analyzing the photographs up there, saving days. I didn't know that. So that was, it was good to learn the importance of real-time satellite imagery and what it can do and how it can change. Yeah. And taking it into, so you don't lose power, you don't lose like your power budget, like you talk about you don't waste time, you don't waste energy and resources by pointing a satellite

to clouds, you know, when you can't see through it knowing that the imaging you're going to get is not good. It would be great if we could apply something like that to how our discussions and our conversations go if we see us going on a ramble zone, only if AI could redirect us, right? I've got a question for you, just for which I'd like to introduce into these post-show VIP. We maybe get a beer and we ask, what would you like to have asked Fenton that now that we didn't ask? There's any questions that immediately spring to mind, which we should have asked him that because it would be nice to know to... The only thing that was swirling around in my brain, and I knew it would be super, like,

wrong side rabbit hole is all I can think about is, you know, this idea of the satellites pointing to Earth and, you know, live Earth intelligence, what would happen if we just went and turned them around the other way? And has anyone done that? Has anyone applied that to astronomy, astrophysics, all of that kind of stuff. I'm sure they have. But it would be really interesting just to shift that perspective outward instead of inward towards us.

Space telescopes. They do that, don't they? Optimization of space telescopes maybe, right? Optimization of that stuff. I'm sure somebody has thought of pretty much everything that they could do. I wanted to ask how thick lower Earth orbit was. I wanted to know, like, how much space all these satellites are playing in. I don't think it's very much space, but I wanted to know that.

And we just have to wonder about it, Mark. Just need to wonder more. Yeah, I will wonder with awe, wonder and awe. Take it for a walk. Well, awesome. Well, there you have it. Friends and Neighbors. This has been another episode of Thinking on Paper.

My name's Jeremy. Mark is with me, as always. We're backstage post-talking to Fenton Buckley from you, It's been a great conversation. We're staying in the space realm. I'm looking forward to finding more space people to talk to. If you know space people, drop us a note. Hello at thinking on paper.x, y, Z.

Check out all of our stuff on our website. We have podcast versions of this. We have YouTube versions of this. If you prefer seeing our faces or you just want to hear us. So this is about you, not about us. Let us know how we're doing. Send us a note. Tell us what you want to see change.

Do you like the hot buttons? Do you like the news section? Do you like this postgame report? Let us know. This is about you, not about us. It's very disruptive. Be curious. Keep thinking on paper.