Latest / UnDocked: The Maritime Transformation Show / Extreme Heat Hits Trade, Ships Generate Green Hydrogen, and AI Content Gets Watermarked
Transcript
- Nick Chubb: there's a little village called Kaub in in Germany. it has a gauging station that is the is the primary gauging station for the for the Rhine. it's it's genuinely as low as as it ever really gets. the Commerzbank is estimating that it will impact German GDP by 0.35 percentage points this quarter if it carries on until mid-September. Hello everyone and welcome to Undocked. Hello Ral Harris, how are you? Raal: Good afternoon, Nick Chubb. I'm very well. I've just got back from the Alps. I've been up in the Alps. Very nice. No, I was on a train at the time. Nick Chubb: lovely. Did you see the eclipse in the Alps? Was it on the p Raal: It was behind mountains, but I don't care. I'm not bothered. Nick Chubb: okay. right, okay. I saw it, I was quite impressed by it. and then I got into the office the next day and everyone was complaining about how rubbish it was. So I maybe I've just got low expectations. Let me tell you a story I saw this week that had me deeply, deeply concerned. It covers two of my favourite things. One is supply chains and the other is parmesan cheese. Raal: Okay. Nick Chubb: I don't know whether you you saw this. Raal: No, I've been in Nick Chubb: But I I Raal: the Alps. I didn't see anything Pressure. Nick Chubb: Right well let me catch you up because this is probably the most important story of the of the year. there are I didn't realise this, but there are five hundred thousand wheels of parmesan cheese that sits in the vault of an Italian bank. and it they're th those wheels of cheese are worth about three hundred million euros. And they have come under significant the the bank's vault has come under significant strain with the hot weather this year. 'cause they have to keep the cheese cool enough while it's maturing. and if it gets too hot it'll spoil. So that sent me down quite a rabbit hole, as you can imagine. Raal: I'm partial to Parmesan cheese but how much did you say it was worth in this vault Nick Chubb: i in the vault they have five hundred thousand wheels at any one time. Those wheels are worth 300 million euros. Yeah. there are about four million wheels made per year. you know to call it Parmesan, I don't want to insult people with my bad Italian accent, but Parmigiano Reggiano, it there is a there is a strict legal definition of what what defines you know genuine Parmesan cheese. They make about four million wheels a year. But what happens is to be able to legally call it Parmesan cheese, it's got to be aged for at least 12 months. So the farmer makes the cheese and then he deposits it at the bank and the in a vault for up to 36 months while it matures, and they lend the farmer 80% of the value of the wheel of cheese. So the farmer solves their cash flow problem. But there is a genuine cheese bank in Italy with vaults full of cheese. Raal: I'm reassured by this. You said this was a concerning story, Nick Chubb: It's cu it's quite an amazing story, Raal: concerning story, but the cheese is imperiled. Nick Chubb: cu it's quite an amazing story, isn't it? well it ha it has been by the hot weather. So there's two reasons. Raal: I know. Nick Chubb: One is just keeping the the cheese vault cool enough has has has the the the use of electricity has spiked by thirty percent in the holding vault. During the kind of peak of the heat over the last few weeks. And you're supposed to keep parmesan cheese between 16 to 18 degrees Celsius while it's maturing. So we're way off. But it gets worse. When the temperature goes above 40 Celsius, apparently cows just lie down and stop eating. They just can't be bothered, which is understandable. And so all the milk yields drop. So you get this double pressure of the cost of maturing the cheese goes up dramatically because you've got to pay for the air conditioning. And then the supply of milk drops. And so in in well, it won't actually impact us for maybe two to three years, but we potentially have a parmesan cheese shortage on our hands. Raal: Getting new technology onto one ship is an achievement. Getting every ship the same benefit is where most digital transformation programs lose momentum. On the AROYA refit, Insight delivered the largest cruise IT transformation in EMEA, on schedule in 11 months across 33 plus work streams, built on a single connected architecture designed to be repeated on the next vessel, not rebuilt from scratch. That's the model Insight and SmartSea bring to Fleet Wide Rollouts. Pilot once, deploy consistently, and cut the time and cost of getting every vessel to the same standard. Because transformation only counts once it reaches the whole fleet. Start your digital transformation journey with Insight. Follow the link in the episode description. Nick Chubb: I've got a couple of other a bit more maritime ramifications if if if you'll indulge me. Raal: Yes you did say supply chain. Nick Chubb: yeah. Well I can't on a maritime transformation podcast, I can't just talk about cheese. So I I did look for what what else is going on in the world that is actually having an impact on the maritime industry. And actually it turns out water levels are becoming a serious problem across across Europe. So there's a little village called Kaub in in Germany. It's just west of Frankfurt. has a gauging station that is the is the primary gauging station for the for the Rhine. and the it the reason for that is it's that it's the shallowest part of the river. So essentially if you're if you're gonna take a goods from the North Sea and through to the kind of industrial heartland of Germany, it's gotta pass through Kalb. And so the depth at Kalb is is really, really important. So it's it's measured every day. The current reading on the meter is fourteen centimeters for today. So it's dropped from two hundred and eight to fourteen centimeters. that's not actually the height of the river. It is the it's the height from a a a datum point beneath the riverbed. so the actual water depth is is something like you add another two meters to it. So something like one point three, one point four meters. But a fully laden five hundred TEU barge draws three meters. And so Raal: Yeah, I was gonna say, don't know much about the brown water industry. I'll say I did look into it quite heavily one time, riverboat cruising, another thing I looked at. But yeah, that's not something you could negotiate very successfully. Nick Chubb: Y exactly. and so what happens is all of these barges are now going through a part load. So if the if the meter is reading 250 centimeters, you can go through no problem with a full load. 500 TU. 135 centimeters it drops to a half load. 75 centimeters, you need four ships to go to get the same cargo through as one ship does in a full load. At 55 centimeters, you need six ships to get the same same amount of cargo through as you would on a on a full load. and no one's published a ladder for what 14 centimeters means. So it's it's genuinely as low as it as it ever as it ever really gets. the Commerzbank is estimating that it will impact German GDP by 0.35 percentage points this quarter if it carries on until mid-September. Isn't that insane? It it Raal: It is, it is. it's, it's, this is the thing we've got to get heads around. As I said, the knock-ons of these things. Do I read a statistic midweek that the heat in the UK this year has cost us, they think 4.4 billion, something like that? I think, yeah, I think that's the, Nick Chubb: Really? Wha in what? GDP? Wow. Raal: I think that's the, estimation that, that I saw there. How, how is the Panama Canal doing? I mean, what's, yeah. Nick Chubb: you asked because that is my my my third one. It's an El Nino year. So you've got the kind of normal well You've got the normal impacts, quote unquote normal impacts of climate change, plus you've got the fact that it's an El Nino and it's looking like 2026 will be a super El Nino, which is like a it's a it's a warming of the Pacific that happens because of a a a current fluctuates. and so you get p particularly warm water in the Pacific, which has led to almost almost no rain at the the basin of of the Panama Canal, the main lake that kind of feeds into it. So back in March, the reservoir was actually the highest level in the canal's recorded history. But then there's been no rain since, effectively. And obviously given that Panama is tropical, you'd expect it to rain every day. So the published design maximum for a Neo-Panamax sorry for the Neo-Panamax lock so the new canal is is 55 feet of draft. it has been cut five times since July and it is currently forecast that by the third of September the maximum draft will be about forty seven and a half feet. So we're taking about three feet, we're taking about a meter off the maximum draft that you can take through the Panama Canal. So roughly speaking, you're talking this is sort of me doing maths on the back of a cigarette packet, but about 875 TEU per crossing if you're if you're to if it's a Neo-Panamax container ship. So you're you're losing nearly a thousand TEU per crossing that you can get through. They haven't cut the number of transits, but they are cutting the drafts of the vessels that can transit. The last time there was a major drought in Panama they actually started cutting the transits down. I if you remember it, I think they do s they do somewhere in the region of sort of 35 to 40 transits a day. I think it went all the way down fifteen transits back in twenty twenty three I think. which then starts to have really major i impact on supply chains. But at the moment, it is it is problematic but not disastrous. Raal: Yeah, yeah, yeah, yeah. But massive, you know, massive impacts for these things as we see down the line as well, particularly if they start cutting the number of transits and it becomes another choke point, doesn't it? Yeah. Nick Chubb: That's exactly it, and that's exactly what the Rhine's become. And and and that that is a pattern that I didn't know much until I started researching this about inland waterways. but it's becoming a kind of a repeated pattern all the way across Europe. so it's it's the it's the Danube as well. yeah, it's it's all over the place. Raal: Yeah, the Danube has it's a big leisure waterway, isn't it, the Danube? But of course it must be a big trade one as well. Nick Chubb: Yeah, it is. It is also a big trade waterway. and also water is used in the water is used for cooling of nuclear power stations and for turbines. Raal: That was what I was searching for. think it's Romania, isn't it? They've had to switch off eight or nine of the power stations, something like that, because they can't. Nick Chubb: Yeah. H Hungary, Romania and France have all had to reduce power generation because of a lack of cooling water and there's a lack of hydropower in Serbia as well. So we've got the the dual problem of everyone's using air conditioning wall to cool down the parmesan cheese, which is critical. but we're also don't have as much electricity, because we're having to to reduce the number of of power plants running. Raal: Well, it sounds like you need a good news story then. On Nick Chubb: yes please. Yes please. Raal: this very subject, I was going to say, know, like, yeah, it's a bit of a shame that we've all kind of, sort of, you know, net zero has become such an unfashionable thing to talk about and shipping is no exception in terms of taking the focus away from that. So I was delighted to read this morning about something that I think is astonishing. Now, the debate around decarbonisation in the shipping world has been for, well, since we started talking about it, was about how can we limit the amount of emissions that ships use and limit the amount of energy, therefore the amount of emissions, and look at creative ways of doing that. Now, so what would you say if I said to you that shipping could actually be the answer for generating? new fuels and electricity. Nick Chubb: Well yeah. Raal: Are you in? Yeah, well, it's pretty mind boggling. It's pretty mind boggling. I've been trying to work it out. This is Drift Energy. And their big idea is essentially using sailing ships to generate green hydrogen Nick Chubb: Okay. Raal: at sea. So not just limiting the amount of energy that's consumed as cargo, literally using ships to generate energy and clean energy, by the way. Now, think about it in terms of a problem with wind turbines. Nick Chubb: Yeah. Raal: Essentially, there are parts of their life where they will be idle because there's no wind and they can't orientate to Nick Chubb: Yeah, yeah. Raal: the wind and that's partly where some of the cost. The other big one, of course, is the infrastructure that you need to be able to harness that wind energy in terms of undersea cabling and of course the installations themselves, putting anything in salty sea water is always going to limit its lifespan for a bit. This innovation is designed to basically do what those wind turbines can't do and go where the wind is and orientate to the winds. And this is basically the vessel undersea turbine generating electricity from its movement. onboard electrolyzers turn that electricity into hydrogen, which can then be taken directly to the port or to a vessel to bunker. And it's a combination of obviously incredible technology, incredible vision, I guess, but the technology itself. there's a big AI component here. Nick Chubb: made Raal: Yeah, all about the routing. So the AI component here, the software element is basically getting the vessels where they need to be and instructing them on the paths that they need to go to be able to generate it. So you're generating clean, green hydrogen, zero impact on the seabed, on the ocean, completely clean. So nothing by way of infrastructure apart from the vessels themselves. And it's more than a concept. So Rena have granted the design approval in principle in May of this year and for the first energy harvesting ship. And they've announced a framework with a commender capital partners to target 50 vessels and million worth of project investment. So it's real, it's happening. Nick Chubb: Wow, that's really cool. So w Raal: It's amazing. Nick Chubb: when would one of these be sailing? Raal: So I'm glad you asked Nick. They plan to launch the first full-scale, I think first energy harvesting ships in next year, 2027. Nick Chubb: That's really cool. Well I guess it makes sense, right? Because if you're moving through the water, I have seen impellers you f on things like, you know, racing yachts and things where they they will they will they will put an impeller under the water line and then that will power some of the onboard systems. But I've never seen them as an idea of actually generating and then storing Raal: Yeah, this is Nick Chubb: and then distributing. I guess you can just send it to whatever whatever it needs to go in the world. Raal: Well, indeed. Now, the other thing, of course, that they're saying here is that obviously because the infrastructure elements of this not being needed, this is a particular interest to sort of island states and things like that. Nick Chubb: Yeah. Raal: So you can put it anywhere. That's the other thing, you know, and most of the alternative fuels, most of them, you know, most of them come with some sort of problem that you've got to overcome and some trade off and, and now the only thing with this is whether the, you know, whether, whether this can be done at a price point that makes sense, I'm guessing, Nick Chubb: Yeah. Raal: because I did, I did have a look at some of the the figures, I think they published them, I didn't make note of them. But you know, the cost per kilowatt or whatever, or kilogram, I'm not sure how you do with hydrogen. and it's quite steep, know, these early prototypes, proof of concept more around the energy itself, but it drops quite quickly, the more of these vessels that are obviously deployed, the more the technology can be scaled. Are you on the website? Nick Chubb: Just yeah, well two things. One is they're an Earthshot prize nominee for twenty twenty five. I'm I'm actually an earth shot assessor, so when maritime Earthshot prize stuff comes up, I get asked to give a view. But actually I didn't Raal: Interesting. Nick Chubb: I didn't get this last year. They didn't send it to me. which is a shame 'cause I quite like it. but Raal: Yeah. Nick Chubb: it says on their on their website they they produce each vessel will produce a hundred and fifty thousand kilograms of Hydrogen per annum and heavily caveat this is a Gemini overview. so this this could well be wildly wrong. but according to Gemini, 150,000 kilograms of hydrogen would power 166,000 homes for a day, or 4500 450 homes sorry, for a full year, which is actually you know pretty good. many homes are there in the UK. Yeah, I d I d I'd it's it's probably not gonna do a big country, but if you take a small you know, a small country like you know, like Marshall Islands or the Caribbean Islands where you've got lots of water around you, and probably relatively modest power requirements, it's it's probably it's probably pretty sensible. Raal: Well, well, they're looking very much at ship operators, which is why this is a squarely in our territory, Nick Chubb: huh. Raal: my friend. So the idea would be that these these could actually bunker your hydrogen green hydrogen powered vessel. where you are. So yeah, being able to do either at a port facility or actually out at sea, think. Nick Chubb: At see. Wow. Very cool. Well cool story. Raal: Yeah, well we all need a little bit of hope don't we? When Nick Chubb: Yeah. Raal: you talk about the fate of cheese and the cows I mean you know these are real problems and what I liked about this story was that it's just like completely changing the paradigm because everything sort of typically we talk about when we think about kind of replacing HFO as the major fuel or whatever. look Nick Chubb: Mm. Raal: to the existing system and think about how it's going to work, know, how are going to replace the oil with something that resembles that as a fuel with the exception of say nuclear. You know, we're looking at like, so the conversation becomes about the poor infrastructure and it becomes about the fuel production and in a very similar way that we're imagining how it moves around stuff. And then you get something like this, which is outlandish and is unproven, I have to say. mean, don't, know, clearly it can work, but you know, at scale and Nick Chubb: Yeah, yeah. Raal: affordably can it, but that to one side, for someone to kind of come up with this as a concept is just really brave, you know. And I looked at the team that they've put together there is quite interesting as well. It's a British founder. Nick Chubb: Ben Medland. Raal: as well. Formerly of Accenture group, which I thought was interesting as well. But the team that they've put together there, it's a real combination of seafaring folk, the kind of people that we regularly kind of feature. There's ex-masters in there, naval architects as you might expect, but also kind of world championship yacht kind of people. hydrogen engineers as you'd expect as well in there. So it's a real kind of, it looks to me to be a really kind of solid, yeah, a solid team to kind of come up with something like this as you'd need it. And then to kind of scale it up, there's a lot of experience there in terms of attracting capital and all that sort of stuff. Nick Chubb: Yeah. Raal: So, you know, watch this space, but I for one would love to see this. I think it would be amazing if, you know, looking out to sea and just seeing these kind of, you know. Nick Chubb: Yeah, it's really cool. Raal: this flow tiller of kind of, you know, energy generating vessels. It's just one further example, ladies and gentlemen, if it were needed that the fate of this planet may well yet depend on good, honest seafaring folk. Nick Chubb: I'm I'm generally speaking, I'm a I'm an optimist when it comes to this stuff and I'm I'm absolutely certain that there is a technological answer to decarbonisation. And we could well have just you know, could well have just found it. If not it, then a contributed to it. We are getting much better as an industry at stealing from other industries, and in particular there I mean yachting. 'Cause obviously BAR technologies, the BAR, who make the wind wings, the BAR stands for Ben Ainsley Racing. So they they they've they they took all of the IP out of the very, very fast yachts that we have been desperately trying to win the America's Cup and failed to win the America's Cup for many, many years. But they took all the IP out of that and have turned that into wind wings, which are now being deployed you know, reasonably widely across across the Bulk fleet. and yeah, so it's it's nice. I I like these I like these kind of stories. Raal: and delivering major energy savings. think the other thing that was interesting to me about it was that this whole kind of, maybe we don't talk about it enough, but sort of ocean economy that's actually there. When you go to sort of investment things, it's a really big potential kind of growth area. from sort of floating kelp farms, that can then generate what we need for biofuels some of the innovative ideas around floating ports and things like that, there's a hell of a lot of stuff going on and probably isn't talked about enough. Nick Chubb: yeah. There's a huge amount of potential kind of locked up there. that I don't think gets talked about enough. Raal: Yeah, yeah, I should say that the tagline I like as well oceans of energy, Nick Chubb: That's nice. Do you where where I thought this was going, when you first started talking about it, it was the the the tidal power people up in up in Scotland. Raal: yeah, that's it. I'm glad you brought that up because I remember reading that and thought, I must must have a look at that for undocked. And then I didn't. So go on. Nick Chubb: They're c it's called it's called orbital marine power. and they've they've been building them in Orkney or or sort of in the waters around Orkney. and essentially it's a kind of vessel that sits on the surface, it then gets anchored to the seabed, so the vessel itself doesn't move. And then it has turbines underneath it, underneath the hull. And it just sits in the tide. And as the tide, because the tide round Orkney has a huge height of tide difference. So I think it goes up and down like six meters. So the tide races out, turns the turbines, then it goes slack for a couple of hours, then the tide races in, turns the turbines. And that the nice thing about tidal power versus versus wind power is it's totally predictable. We know exactly what the height of the tide's gonna be at two in the morning in 2038 because it's all based on what the moon's doing. and so you don't have all of this problem of of not being able to predict it. The problem Unfortunately for the Orkneys, as I understand it, is actually that th these this system is actually too efficient and and Orkney ends up with far too much power and they don't know where to store it. So they actually produce hydrogen on Orkney and now all the buses are run off hydrogen because of these tidal projects, because they've they've they don't yet have a way to send it send the power back to the mainland. And they've got too much. Raal: Yeah, yeah. And this is one of the things I really liked about the Drift Energy thing as well, that it was actually kind of generating the green hydrogen actually on board. We've seen that with some of the carbon capture solutions as well, haven't we? Actually sort of onboard processing plants as well, which solves a huge amount of problems, it? Nick Chubb: Yeah, there's a there's an interesting company. I can't remember the name of them unfortunately, but I w otherwise I would give them a a shout out. they have found a way to make In instead just capturing carbon as a waste product, they found a way to scrub it and process it on board, particularly when it comes from LNG as a as a feedstock. so if you've got an LNG fueled ship and turn it into pure graphite. and so it basically becomes like a at a grade, instead of it just being like a CO2 capture and storage, they actually get near enough kind of pure carbon that can go straight back into the supply chain because it's got all sorts of chemical uses, and so you get this really nice Raal: Mm. Nick Chubb: kind of circular economy. economy bit to it. So instead Raal: Yeah. Nick Chubb: of having a a waste CO2 to to discharge when you get to port, you get a a helpful yeah graphite essentially that that that comes off the ship that can then be used in lots of industrial processes that has economic value as well. So it's not just the ship having to pay to get rid of something. They can actually sell a byproduct of the of of of the emissions. Raal: Yeah. And these, you know, these are the kinds of innovations that are so important when we're in this sort of doom loop that we sometimes find ourselves in and looking for a way out. And then, you know, I'm glad you're an optimist. tend to be a little bit more pessimistic, but mainly so that I can, you know. I'm pleasantly surprised when something good comes along. Yeah, exactly. Yeah, I think so. Nick Chubb: Low expectations. It's a key it's a key it's key to a happy life. Tal talking of innovators, have you seen the the Singapore Maritime Authority has announced I guess a a kind of th they've sort of published a take stock of of their P seventy one startup program? Raal: this is the startup investment thing. Nick Chubb: Yeah, it's their it's it's their accelerator, their maritime startup accelerators. Obviously maritime is hugely important to the economy of Singapore. I think it makes up something like seven percent of G G D P and so it's like strategically important to the government. and so they created the Maritime Report Authority of Singapore created a program, I wanna say in like twenty eighteen, that helped to accelerate maritime startups in Singapore. It's called Pier seventy one. they've done an eight year announcement, I guess. that they have had more than a hundred and seventy startups go through the program now. thirty technology solutions have been deployed to the industry. through that program and a hundred and thirty million Singapore dollars has been invested from venture capital investors through the program as well. so it's like a really, really nice story of a you know kind of genuine impact of a of a government backed accelerator, which I can tell you is very, very rare. Raal: Yeah, interesting. Any names there that we would recognise? Nick Chubb: alpha auri who Raal: yeah. Nick Chubb: were acquired by zero north they they provided high frequ high frequency data for for ships river I don't know if you know River but they're they I think they've moved on now but they were doing a kind of seafarer identification cruise certificates type platform. I think they've now moved more into the identification space. eye gauge, they've done quite well. very very clever approach to digitalization so like a lot of like Japanese ships for example Japanese built ships I used to work on a Japanese built ship it's all analog dials right so there's no digital dials for or or at least historically you know if you if you wind the clock back probably just five ten years you know, like the rev counter on the bridge would be an analog dial, and the oil temperature meter in the UCR would be an analog dial. And so when it comes to actually kind of getting digital data in a high frequency way off the ship, it's really, really difficult because everything's measured with these kind of these dials that sort of move up and down. So what they did is they just created a set of cameras that sit over the engine control or bridge control dashboards and just look at these dials. And then they used AI and machine learning to basically translate analog readings into digital readings. And then voila, you've got high frequency data collection that comes straight off the ship instead of having to go in and fit all sorts of really fancy sensors that, you know, that that that that cost tens of thousands of dollars. So really, really cool company. Raal: very interesting. Is there a reason that the Japanese persisted with the analog dial? Nick Chubb: can probably take us off on a right tangent here. we we we actually we did research on Japan's innovation culture quite a long time ago. This is like back in twenty twenty. and we found that there is and apologies to any of my Japanese colleagues, if I if I paraphrase this slightly wrong, but we we found that there is a a general philosophy around using kind of withered technologies in sort of new and innovative ways. so a perfect example is if you take Nintendo. so you you've got do you remember the Atari Lynx? Just about? Yeah. Okay. Raal: Just about, just about. Nick Chubb: Well it it came out at about the same time as the Nintendo Game Game Boy. And the Atari Lynx had a sixteen bit custom processor on it versus the Game Boy's eight-bit custom processor. the Atari Lynx had a three and a half inch color L C D screen, and the n the Nintendo Game Boy only had a two point six inch, and if you remember it was like monochrome, black and white. The Atari Lynx had four channels of stereo aud audio. the Game Boy had just a single single speaker. and the Lynx had six AA batteries and you got about four hours of gameplay from it. And the Game Boy needed four A batteries, but you got thirty hours of gameplay. And it turns out that actually back in back in the 90s, and I think the same is true today, that no one really cares about the quality of the gameplay. They just want good gameplay and they want it to last for a long, long time. And so the the Lynx sold three million units, and the Game Boy sold 118 million units. And that's because when Nintendo were putting it together, they basically took much older components, much less cutting-edge technology. And put it together in a way that was much, much more simple, straightforward to manufacture. Because of the poor computing power as well, it was much, much cheaper for games designers to manuf to to create games and manufacture the cassettes and all this sort of stuff. And so it ended up becoming this kind of runaway success. Raal: Hmm. Nick Chubb: And everyone kind of forgot about the Atari links. So there is a general philosophy in in sort of Japanese technology culture that actually you can use well established technologies in new ways and often you'll you'll get a better better outcome than trying to do something that's massively on the on the cutting edge. so there's that which shout out in the comments if you think I'm talking rubbish. and there's also the fact that Japanese ships are just very well looked after and so they last a lot longer. So you've you've you've got you know kind of an older profile of ship. And so you might have ships sailing today that were built in like 2000 to 2005. And so the cost of upgrading those ships into sort of totally new digital infrastructure, putting new sensors in, all that sort of stuff, versus just sticking the cameras on board, is yeah, like orders of magnitude different from each other. Raal: I do remember discussing with an engineer once who was lamenting digital displays. because he said when he went to the engine room with analogs, he could kind of look across the landscape of dials in the engine Nick Chubb: Yeah. Raal: room and just like go, yeah, it's okay. Everything's where it needs to be. Whereas Nick Chubb: Yeah, yeah. Raal: when it came to looking at digital reading, he had to look and process each of the digital readings. And I wondered if that's kind of like, you know, one of the, one of those things we don't always think about. We assume digital is more accurate, therefore it's better, but it- but it doesn't necessarily follow that it's more holistically, you know, a Nick Chubb: Yeah. Raal: better thing. And so that was what was playing on my mind there, where the Japanese were kind of like making a conscious choice to stick with. Nick Chubb: Yeah, yeah, no, I I I think that's absolutely true. I think that Nintendo Game Boy examples are very, very similar. it's the same with cars. You know, the display on my car is awful it's all digital. There's no buttons anymore. changing the temperature of the air con takes about three hours. Raal: it does strike me as slightly bizarre. that you can't use your mobile phone whilst driving, but you've got like an iPad on the back, which is like you said with Nick Chubb: Yeah, yeah, yeah. Yeah. Raal: the air con and stuff like that. mean, how can that be safe, you know, when the mobile phone is not, but you know, there you go. Nick Chubb: let me give you some more alumni of Pier seventy one 'cause actually there's there's there's quite a lot of yeah, we got we got sidetracked there. Raal: Yes, back to the story. I hope you Nick Chubb: Bound for Blue, who we've talked about quite a lot on the on the podcast. they Raal: yeah! Yeah! Nick Chubb: they went through Pier 71. Raal: Wow. Nick Chubb: Shipping Systems, who just raised, I think it's fifty million dollars. went through the program. Daphne Technologies, who are in carbon capture. Raal: It's gonna be good. Nick Chubb: Sideome Security, so there's a there's a fair few names. Hang on. there's definitely a couple more. Yeah, there's a fair Raal: Yeah, some successes. Well, you know... Nick Chubb: few successes there. Raal: Yeah, mean, Singapore, obviously, you like you mentioned there, how important the economy is, but we'll be talking about sort of the broader, you know, decarbonisation space as well. They were one of the very first people to step into that, weren't they, and to start sort of resourcing solutions and so on to it. So we need these people, we need them. Nick Chubb: I think it's really important to say as well, and speaking from experience, I I was in a in a previous life a policy fellow at the at the sort of public sector management consultancy company public and we were looking at enabling startup driven innovation through public policy and particularly in the trade space it was all during the time around Brexit and all that sort of stuff. So the every the government was very keen on working out how startup innovation could help with the political try not to swear here, shambles, that was being created through through Brexit. and anytime the the government and civil service wheels start to turn anywhere near a startup, it very, very rarely ends well. b because they are generally speaking just turning at two very, very, very different speeds. and so However PS 71 functions, to have a an alumni list like that and have it backed by the Singaporean government and have it effectively be part of a government agency is genuinely very impressive because I I have don't think I have another example anywhere else in the world of of kind of public sector driven innovation that's been so effective. Raal: the failure rate is pretty astonishing, isn't it, with sort of, you know, investing in startups, it's And of course, you know, it's a numbers game and you might get lucky, it's a very precarious thing to do. And I know you've got some strong feelings, I think, particularly in the UK at being quite good at sort of funding research, but not really very good at helping that Nick Chubb: Yeah. Raal: translate into business success. Am I right? Nick Chubb: you you are right. Yeah, that's good memory 'cause I've probably r ranted about this in a pub to you at some point over the years. Raal: Yeah. Nick Chubb: yeah, in the UK, we have programs like UK Shore, we have Mar Marie UK, which is a maritime research and innovation center and and they're really good at getting things from so we going into kind of innovation jargon, something called TRL, which is technology readiness level. And it goes from technology readiness technology readiness level one, which is like very, very kind of basic idea that's that's just sort of been proven in a lab, to technology readiness level seven, which is like commercially operating technology that th that there is, you know, there's a business model around and and and you can actually make it work as a as a business. We're really good in the UK at sort going from level one to level four. And then there's this sort of gap between level four and level seven where we don't really have the funding mechanisms for. I think programs like Mario UK are supposed to close that gap. personally, I'm not really convinced that they do. and the venture ecosystem in the UK is also, you know, it is it is limited. it's it's probably one of the best in Europe, and particularly London, sort of one of the best sort of hub cities in in Europe or for raising venture capital. But compare it to the US, and actually to be totally honest, compare it now these days, compare London to Singapore, and I'm I'm not actually sure London really touches Singapore in terms of the quality of investors in you know on a kind of per square kilometer basis. yeah. So I I think we do have a genuinely problematic gap there. and we end up with a lot of we end up with a lot of research programs that either just sort of collect endless grant funding and just go round and round in circles. also end up with quite a lot of I think the the the the term that gets used is sort of zombie companies where they effectively end up just sort of reliant on the next grant and they never just quite manage to cross the line into Raal: Yeah. Yeah. Nick Chubb: having something commercial. Raal: Yes, yeah, I've certainly seen that myself with a few EUs seem to be good at that. Nick Chubb: the EU's amaz yeah, amazing at it. the Horizon programme, I Raal: Yeah. Nick Chubb: mean you know, you're talking Raal: Yeah, there's a few. Nick Chubb: tens to probably tens of billions at this point of funding that's gone into stuff over over decades that I'm I'm not actually convinced you'll find any really important European success stories in. Raal: But if you are interesting, if that's your thing, particularly in maritime, I draw your attention to the episode we did with Stylianos of Lomar Labs, Nick Chubb: Yes. Raal: because that's very much a problem that they're looking to help with. So as a ship owner with a sort of innovation arm, where they actually kind of look at helping companies to... to get their solutions up and running and certainly looking at some of that commercial aspect of it as well. So not just from an R &D point of view. And he talks very much in that around the difference between R &D and innovation in terms of Nick Chubb: Yeah. Raal: making viable companies. Nick Chubb: Yeah, he that's actually that's a really good episode actually. It's there's lots of valuable advice in there. Raal: Now, of the things we talked about the EU there, we talked about enabling innovation. So I think it's probably came up last week, The EU-AI-EU Act. And Nick Chubb: Mm. Raal: we talked about that in the context of MINTRA had just released a training course and it caught my attention. But I hadn't really looked at the actual legislation itself. you depending on how you look at it, you can either take the view that the EU has been very, you know, proactive in trying to protect people from the potential downsides of AI, or you can look at it as a limiter of innovation, but with the EU act, the EU commission is effectively, it is a code of code of practice on transparency of AI content and This is the first kind of, I think it's pretty much the first constraint, isn't it, on the explosion of AI. There's been a couple of private things around copyright and stuff like that, but I think on a governmental level. So it's interesting that this has happened and it's been signed up to by, I think, all of the major ones. So the one that's in the news at the moment has been Anthropic because Claude, has announced that they've implemented this watermark. So we'll talk about that in a second. But all of the others, Google, OpenAI have signed up to this. Claude has been the first one that's talked about how they're implementing it. But essentially, this is to ensure that there is transparency around where and how AI is used. Obviously, one of the things that they're very keen on doing is anything that that sort of might sort of enter the political arena, I guess, in terms of fake news information, that kind of stuff. So that's a a big driver to it as well. So the idea is that essentially that there are obligations on on people developing AI tools, and also those that are using those tools. professionally, should say, importantly, that they're very transparent about the fact that this is being used and that people are able to see it. And it's caused a bit of a furore in the AI world. I Claude's had quite a lot of people getting in touch with him. And so just to give you a little bit more background, it's working on two levels. So one of those is actually the detection of it in text itself and I'll explain that in how that's going to work or how they've got it working certainly as far as Claude is concerned. And the second one is a sort of metadata approach. So when there's an image video generated that within the metadata of the image it's declaring effectively how it was generated and all that sort of stuff. But there are a number of obligations, well, on people which they may not be aware of around sort of where you need to declare. So for example, if you use an AI chatbot, the law is now that you have to declare that this is an AI chatbot. Quite strangely, as I get out of jail or cubbing, unless it was obvious to a person that that's what it is or something, I don't know quite how they... how they go about doing that. I presume if you've got like a bouncing robot avatar or something, then they kind of, you know, maybe that's obvious to people. don't know. But you can't like, for example, sort of say, well, I would like to pretend that the that you're speaking to a real, I don't know, real employee when in actual fact, it's not one that kind of stuff. So the interesting thing there now, obviously, putting, you know, watermarks in metadata and stuff like that, I think most people can sort of understand how how that works, that it's machine readable. slightly concerned that that sort of stuff is typically pretty easy to sort of take out. But most people can imagine sort of what that looks like. The thing that's slightly more tricky is this watermark in text, because this survives the copying and pasting of that text into some other form, another document, you know. So this is not the text. Yeah, well. Nick Chubb: Knowledge workers everywhere just just just shuddered. What? It survives copy and text. no. Raal: Well, it's important to think about this. Yes, and not only, not only, because the assumption was that when you talk about a watermarking text, people were sort of assuming that there's like some characters in there. Those of you that worked on websites and in any kind of search engine optimization, there was this brief period where people would use white text on a white page, for example, to obfuscate some text. Nick Chubb: Yeah, yeah. So common prompt injection technique. Yeah. Raal: The text was horrible, but very search engine optimized. So rather than ruin the text of your carefully crafted website, you'd have sort of hidden text effect fundamentally. And then people were kind of with this watermark were assuming that there was some sort of characters, you know, sandwiched between the text. That's not how this works. This works in a completely different way. And it's about the actual choice of words. So particularly, Nick Chubb: Interesting. Raal: Yeah, and the way in which the words now you've talked about how AI sort of generates text. And that's sort of very interesting in terms of the way in which it chooses the choice of words and it's probabilistic, know, what word should follow which word to kind of end up with something coherent. So this is actually affecting that choice to be able to make it readable and detectable. And so, so there is a potential to influence the quality potentially. You could say that you could write any sentence, you know, a number of different ways, and you may not necessarily notice. one of the areas that this is getting people very upset though, is it is in the generation of code because code is effectively text. So, so Nick Chubb: Makes sense. Yeah. Raal: it's actually influencing the structure of that code to, to make it survive, you know, survivable from a copying Nick Chubb: Yeah. Raal: and pasting point of view. And I mentioned the other. Other people that have signed up to this I understand one of the things that I don't know how they did it But I believe Google have actually been doing this for a while So so, you know shame shame on for claw that they're the ones who've copped it for implementing this And and some other interesting things in there as well. Like you mentioned there a lot of people would be frantically thinking now. Well, how am I gonna Obsticate, know this this code. Now, one of the other things that we should say in my, my look through it is that even if you generate that text yourself, so if you write long form and you use AI to condense it, which is something I do a lot because I overwrite everything, I over talk everything. I also overwrite Nick Chubb: Yeah, yeah. No. Raal: everything. So I'm quite often struggling to get things down through economy. Nick Chubb: Yeah. Raal: The minute that you do that, it's then announced as AI generated. it's sort of, Nick Chubb: Interesting. Raal: you like the idea, it's contaminated at that point from an AI Nick Chubb: Yeah, yeah. Raal: perspective, which is an interesting thing to work with. Also, I mentioned that, and I know you're a big fan of this as well, these kind of Claude workflows where you can generate nicely formatted documents and PowerPoints and all those sorts of things. You may start with your own text, put that in to generate. something in a more branded format or whatever with one of your constructed skills in Claude at that point, you also then have that marked as AI generated. Now, just to make a really important point here, the way that this works from an EU perspective, it's a little bit like GDPR. So what you see within the way that they've set this up is there are different roles. So you've got to a provider. So a provider is anyone that's kind of building out the AI software. And that obviously includes Anthropic and OpenAI and those kinds of Nick Chubb: Yeah. Raal: people. But I would imagine that also anybody that's generating software using similar things for more of a more general application, not just as a chat bot. Nick Chubb: Like a C C R provider, for example, that's now bringing AI into the product will have to comply. Raal: Yes. And then the deployer. So the deployer is most of us, people who are using AI Nick Chubb: Yeah. Raal: systems professionally. So if you create something daft in chat, you're under no obligation to do anything. to put on your Instagram feed or whatever, that's your business as far as they're concerned. But if you're working for a company and you are using AI tools, the company that you're working for is in this case, the deployer. so there are obligations on both sides. And if you are railing against Claude for putting this watermark in, the actual counter argument to that is that what they are doing is fulfilling their obligations to enable you to not fall foul of the law. Because you now have obligations as the deployer to to follow these. And it's interesting, hats off to Mintra because when we were talking about this the other day, I mean, I was looking at this about like, companies are rolling out AI solutions and there's obligations in there, but I hadn't twigged to putting these two things together. So I took a bit more of a deep dive in it and there is stuff that could catch you out, particularly, let's say that you found and there's already going around on the internet at the moment. various people publishing ways, or here's a skill you can put into Claude to remove the watermark. Well, if you do that, you are breaking your obligations as the, the deployer to make it detectable. Nick Chubb: think it's worth it's worth saying for anyone who is a deployer, to you to use the proper parlance. Like general good practice would be that It that's kinda ch I I see as kinda two two tests to take, right? So number one is would someone feel misled if they later found out something you produced was AI generated? And if this if the answer to that is yes, you should absolutely make a very plain declaration, not encoded through a watermark like it should be it should be clearly shown to be AI generated. And then the second test is like would someone rely on the app put differently if I knew it was AI generated rather than generated by by a human? I think if the answer to either of those is yes, you should actually just have it very clearly like visible to a human on a document or where you know whatever it is that you've produced. because the interesting thing about this clawed watermark of I've done some panics reading about it, is that y there isn't it is although the watermark now exists There isn't actually yet a way to to check anything against it. They haven't actually launched the API that will allow you to to check if if if the if if the if the work's been used by Claude. So that is coming, I imagine. Raal: Well, yeah, yes. And some interesting things around that as well. interesting point is your one there, Nick, which is like, know, where do we fall on the transparency thing? Because are we saying that the concern here is that people will find out that you've been using AI? Well, if you're the quality of the work is acceptable and you've managed to generate it in less time and ultimately therefore save them money somewhere along the line. Why should they care? People do though care about this and certainly people feel around a personal thing around human creativity and that something has more resonance. They don't wanna feel like they've been cheated. It's a potential pathway to that. I think the bigger thing, mean, certainly when it comes to images and things like that, the fact that, you know, it's getting to the point where you can't really trust your eyes and think about like, you know, Nick Chubb: Yeah. Raal: But there is another interesting thing in this, which is a side benefit to the AI companies. And that is that the proliferation of AI content has meant that all these AI systems that are training on human content have started to run the risk of training on AI generated content. And a friend of mine gave the expression which I rather liked which was pap will eat itself. Which is the after the the pop band pop will eat itself. So yeah the idea that you know the slop trains worse and worse and the slop gets worse and worse and worse. So to your point when they when these are published these standards are published in some format it allows those systems that want to train on human beings to ignore or filter out potentially a lot of content that has been generated by AI. But I think the point still stands, know, do we take from this that in some way that it's inferior quality if it's been touched by AI or do we believe that if it's used in the right way and judiciously, you actually get a better output with AI? And certainly to my point around it being able to reduce voluminous text, know, it's certainly in my case saved people from reading several pages, you Nick Chubb: Yeah. Yeah. Raal: know, by working with it to kind of reduce it. But I'm still there and I'm still, you know, making sure that the important points are still represented. Nick Chubb: Think I think if if if your work is is the quote unquote knowledge work, I I do genuinely think in ten years time working without AI will be will be kind of the same as working without a computer in the kind of early noughties. you know, still possible but becoming increasingly irrelevant. and so I think it all comes down to the quality of the output, regardless of how it was produced. But I think those disclosure tests are really important, around yeah, w if if I mean personally, but I I don't know, I I spend a lot of my time thinking about words. but I I it reads incredibly obvious t to me when people are have used or are using various different AI tools. You can you can almost work out which AI tool was used to to draft the post or whatever it is that you're reading or the email. Raal: Yeah, yeah, yeah, yeah. Nick Chubb: and but but but put parking that to one side, I think the quality of the output becomes much, much more important. bec particularly as the cost to produce it goes near enough to zero. The cost to read it, the cost of of my brain power and my time and my effort to read, for example, my emails, reb it I have lip you know there is limited bandwidth there, whereas the the cost to actually send something to my inbox ha as it's dropping to zero means that there's now a massive weight in the other direct the other direction. and so yeah it all c to me it all comes down to to the quality of what you're putting out. and then putting sensible disclosures in place. And I think if you're relying too heavily on a clawed watermark or and it looks like all of the other major shops have signed up to a to to a kind of a a set of ethical standards around this. So they'll all be out producing something like this soon. I yeah, I I I I would suggest the prudent thing to do is not rely on those watermarks as a way of finding out. Raal: Well, let me ask you this. You said all of the houses. All but one. Can you guess who? Nick Chubb: Ooh. Deep seek? I don't know. Raal: No, it's Nick Chubb: Okay. Raal: Elon Musk. Yes, Nick Chubb: Grog. Yeah. Raal: he's likely refused and said start off, which is interesting because as I said, it's completely predictable that he would do this. But yeah, as I understand, they aren't going to sign up for it. But this is interesting because a bit like some of the maritime laws that we've seen, this one very similarly, even though you're not a business in the EU, it affects you if you wish to have any kind of distribution in the EU. So Nick Chubb: Yeah. Raal: although this is an EU law, this applies to anyone unless they want to pull any of the licenses for for use in Europe, they're going to have to deal with it. Now, I guess Musk, if he's refused to sign up for it, is of the opinion that he doesn't need the business there, Nick Chubb: sorry, just to clarify though, that code of ethics is is different to the to the law. so I don't I don't think although there are disclosure requirements in the law, doing it through watermarking like this, you i is not a hundred percent necessary. There there is a separate hang on. There is a separate code of practice and transparency of AI generated content that has a hundred and ninety signatories on it that is that is different to the act as I understand it. Raal: time to look at it. Nick Chubb: It's a it's a code of practice, which a hundred and ninety organisations have signed up to. Groc not being one of them. But that is different to the to the to the EU AI Act. Raal: Okay, okay fine. Good. Nick Chubb: But you're right, probably if anyone's gonna sail close to the wind on this, he will. So the other thing that to that's worth thinking about is, you know, we are rapidly moving towards you know, on on device local capability for you know, so I I can see a few I can see a future where your next generation of MacBook will have you know, a local set of weights that are tune you a l its own large language model tuned to a local set of weights that is tuned to you as a user. And so, you know, the idea of kind of pinging into Claude to get a response sort of becomes more moot then. But and that's where I think, you know, th that's why why I say I I think that, you know, the the i it will be like turning up to work without a laptop, you know, if you're if you're trying to turn up to work without AI in in ten years' time. Raal: Well, you've got to look at this in the history of things. you look at like a graphic designer used to be able to have to do calligraphy, type setting, all Nick Chubb: Yeah, yeah, yeah. Yeah. Raal: of those things. You think about the skill involved in the early days of Photoshop and what people had to learn to be able to do things that now can be achieved in the field. So the one thing that doesn't Nick Chubb: So so actually so someone described me the other day as as using re refusing to engage with clawed design as a graphic designer would be like refusing to engage with Figma or Adobe, you know, ten, fifteen years ago. And I think that's a good way to to think about it. It is a new set of tools that allow you to work differently, rather than a an outright replacement. Raal: Yeah, and this is where the output becomes really important, right? Because when you look at sort of slop that's generated, yes, okay, it's a little bit better than it was, it's getting better, but you can still see the lack of artistry to it. Whereas, you know, if you're working with it and it's your sort of your taste and your scrutiny and your determination to get something to a quality, you know, level, to a level of quality that... that's going to do what the job of what you're working on. I you said there about the output. That's the point really. What's it that you're building it for? And I believe very strongly that you can have all the tools and automation in the world, but if you don't have that vision and you don't have that taste to discernment and credibility to be able to assess the output and not to stop working on it until it's reached the level of quality that's needed to do the job. then yeah, it's, it's, it's, seems foolish. Nick Chubb: it does come back there in particular to quality of content, right? Because I I I got sent a a pitch the other day that was like This is a platform that McKinsey consultants use to do their slide decks now, you know, instead of because they always used to just sort of outsource it to India and then have people create their slide decks overnight. Now the McKinsey consultant sends it to this platform and I was like, well that's interesting. We do a lot of slide decks being a research company. I'll try it. So I sent a report and said, we'll turn that into a slide deck. the slide deck was utter drivel. Like it was just total meaning. It was like fifteen slides. It looked lovely. Great design. Just meaningless crap though. You know, but and and that was I hate to say it, it was pulled out of our research. I was like, I I would never have gone Raal: Yeah. Nick Chubb: in that direction, you know. So I so I think there is that there has to be some caution with it. Just 'cause it looks nice and it looks polished. Doesn't necessarily mean it's good. and I think Raal: Yeah, no totally. Nick Chubb: AI is bad at discerning good and bad. Raal: Yeah, absolutely. And the other thing as well is if, you know, again, it speaks to the quality control and how you do it, because any experiment that I've done with creating depth, I'd say I've had to put another 25 % of the effort in from the beginning to kind of fix various issues with it. Nick Chubb: Yeah, yeah. Raal: And one of which is, I mean, I can't rely on it at all for creating visuals because they can look really, really effective, but actually the errors, maybe core design is better. haven't, I haven't experimented with that, but, you know, what it does in terms of making things look nice and symmetrical sometimes hampers and, and, know, Nick Chubb: Yeah, yeah. Yeah. Raal: gives, I would say, depends on the context, I guess, but it kind of potentially false information because it looks nicer and more balanced, you know, like, so you really need Nick Chubb: Yeah, yeah. Raal: to know, which again, it just speaks to that whole point around having that quality control. bit in there and that's that human oversight that's really important and I think human oversight may actually be one of the things in the EU Act actually. Nick Chubb: Yes, depending on the risk category, yeah. Yeah. there's Raal: Yeah, yeah. Nick Chubb: a there's a it's the next tier up from low risk. You you do need human oversight. and then the tier up from that you need human oversight that is in the EU physically. So you have to have people on the ground in the EU physically. One one f final point as a way of wrapping up on human oversight. producer Tom sent us a very funny post earlier. a a number of chemistry research papers about kidney failure have popped up using the term kidney disappointment, because they've been reworded by AI without the right context. And so there's lots of cases of kidney disappointment going on. Raal: Wow, well, may you never suffer from kidney disappointment. Nick Chubb: Indeed, indeed. Well on that happy note. Thanks, Raoul. Raal: See you next time. Nick Chubb: Bye.