Tapping onto AI to build a more sustainable future with Recursive AI
25:53 Share

Hi, everyone. I'm Theo Davis, Regional Enablement Leader at Google Cloud in JPEG. Thanks for listening. And my name is Paris Tran. I'm a Territory Account Manager based right here in Singapore. And we have got such a good one today, Paris, because obviously, unless you've been hiding under a rock, you have heard AI flying across left, right, and center every different space when it comes to work, even the home. And we have recursive AI today who have merged all of the

or really found an intersection between sustainability and AI? Yeah, I was about to say it's the best of both worlds. So we have AI and sustainability. So I'm not sure, you know, what's the statistics right now, but these are the two words that I heard a lot these days. With that, let us absolutely get into it and race away here into the green future. Here we go. I cannot wait.

Welcome everyone to the show and especially our guest Tiago, CEO and co-founder of Recursive AI. Tiago, tell us about yourself. That name sounds, is it Brazilian or Portuguese? And we'd love to hear a bit about your journey and as well as that, what was the latest digital download that you made?

Thanks, Theo. Happy to be on the show. Yeah, I'm originally from Portugal. Grew up in Lisbon, Portugal, but I left the country very early on. I was always eager to explore and discover new places. So I spent some time in Germany and the UK as well before landing in Tokyo, where I've been for the last...

four and a half years. As for digital download, I think the last thing I've got recently was a book for my Kindle, Scaling People, which is a book from the ex-CEO of Stripe, I believe, about scaling organizations, which is very interesting for me at this time since Recursive is going through a period of growth and I'm trying to figure out how to manage all the teams. So it's very interesting reading. Brilliant. Hajime mashite. And how's your Japanese actually since you've been in Tokyo for a while?

It's a slow going. So Japanese is a very hard language to pick up. But recently, actually, I've started learning a bit more because my Japanese was limited to restaurant Japanese only. But now I feel like I should step up to business level Japanese. But it's going to take a while. Thanks for sharing. Yeah, that's fantastic. And hopefully the next time we get to speak with each other, you can tell us a little bit more about the summary of the book that you just downloaded and do that in Japanese.

extra chat. The gauntlet has been set now, Tiago. Am I, guys? Am I?

Yes, so fantastic to have you here with us today, Tiago. Maybe you can start by telling us a little bit about Recursive. How did your company come about? And maybe sharing a little bit of your mission. Yeah, thank you so much. So Recursive started around two and a half years ago, and I was already in Tokyo, Japan working for a startup. While working for that startup, I was starting to feel like I want something a bit more out of my work beyond just, you

you know, having a job, right? And so I talked to someone who used to be part of that startup where I used to work, Cogent Labs, Katsutoshi Yamada. He's the co-founder of Recursive. And we started having some informal chats. And actually, I just briefly mentioned, you know, I'm looking for some kind of impact in my work.

And during the course of those conversations, he actually started to talk about sustainability. And we found some kind of like good connection on those topics because I was talking to him about how I feel like Japan is wasting too much plastic. So I don't know if you guys have been to Japan, but the konbinis here, all the supermarkets, everything's wrapped in plastic.

And as someone coming from Europe, I was a bit like, maybe there's a better way to do this, right? So I was ranting a bit and he was like, it seems like you're really passionate about sustainability. Katz Tushy also has a background in sustainable fashion. So he's a serial entrepreneur. So we kind of bonded on that. And because my background is as a research engineer in Google DeepMind, which is one of Google's AI research laboratories, we thought like, why not merge that

skill with our passion. And that's where the mission of the company came about, which is to solve sustainability problems with the use of AI technology. Brilliant. And of course, we all know, the three of us, but maybe for the audience, a new term is Zoogler. And you are in fact a Zoogler, Tiago, being an ex-Googler.

So great to have you previously with us and hear about what you're doing now. And it's interesting about the plastic you mentioned in Japan, because I also believe whilst everything is a lot of plastic and wrappers, there are also hardly any bins anywhere. Like you have to bring your plastic wrappers and other waste with you and find somewhere to put it, which you can't find on the bins on the street that you would normally find in a large city, right?

That's true. I tell all my guests coming to visit in Japan that that's part of the quintessential Japanese experience, carrying your trash around on your hands. So I try to sell it as an experience, a tourism experience here. Love that. Maybe your merging of AI and sustainability can also help to solve that challenge. But it is in general a super unique concept, obviously one that's so relevant for our world now and everything that's been happening lately.

with the AI space. Our Google CEO recently talked about how this is even bigger than perhaps the invention of fire, the wheel and the internet even. I wonder in general though, how are you harnessing the power of AI to drive sustainability? Perhaps in a bit more detail there.

Right. So that was one of the key conversations that we had when we started Recursive was both me and Toshi had this idea that, you know, AI is going to be a fundamental technology in the future. And, you know, even though it goes up and down, it comes in waves. Sometimes the hype is higher. Right now we're in the high hype phase. Maybe two years ago, we were a bit low hype, but I think it's going to keep going. But if you actually look at the trend, it's getting more and more popular.

useful and the practical uses are increasing, right? So what we've observed is that throughout history, it seems like most problems that we have developed as a civilization tend to be solved not by trying to revert back to the Stone Age or revert back to not having those things that

we are used to, such as healthcare or transportation or all the conveniences of modern society, they tend to be solved by innovating such that those things become better, those things become more efficient and so on. And we believe that AI is one key technology there because it can help

make processes more efficient. It can help us improve R&D to discover new materials, new drugs, new ways of doing things. And it can also help educate people, improve their work, and it can even help prevent certain disasters that can come about due to things like climate change or social media, for example. That's also a big issue where more on the societal level, we can try to prevent and mitigate those issues there.

Well, that's fantastic. I'm just wondering now, since you mentioned AI is like the fundamental of how life is going forward and it is, as a matter of fact, Google also play a bigger part in the AI game right now. And I'm just wondering how the AI plays a role in different pillars that you have in recursive. Maybe you can share a bit more on that.

Right. So, yeah, I think I already touched a bit on those areas, but we actually formalized it. So we have a white paper in our website, but we formalized the ways in which AI can be used for sustainability. So one of the key issues in the very beginning was, as you guys have said, right, it sounds very niche, like AI for sustainability. A lot of the people we were talking to were like, what is that even, right? Like AI is just the thing to like detect cats and dogs, right? At least at the time, that's what it was. And so...

It was very puzzling for many people. And so we've tried to formalize it, right? So I've actually spent the first six months of Recursive mostly thinking about this. And what I came up with was the system of the four main pillars where AI can be used for sustainability.

And so we've identified accelerated innovation, that means to help research and development. And I can cover a bit more on what that means. Higher productivity, better education and work, and prevention and mitigation as the four core pillars of where AI can be used for sustainability.

And so those are a bit abstract. Let me give you examples of each. So for example, accelerated innovation, what we're seeing is the use of neural networks, which are one of the core pieces of AI infrastructure. Neural networks can be used to simulate physical systems, right? So for example, they can be used to simulate turbines in engines. And we can do those simulations thousands of times faster. And so that means we can find more efficient mechanisms with less material

better fuel consumption and so on. With higher productivity, I mean, that's what most people tend to be using AI for. So that means automating manual tasks, which I think maybe we can talk about that later. But I think that's very important, not just for sustainability, but also for enabling humans to come up with creative new solutions, right? Because if we embrace that thesis, that innovation is the key,

to sustainability, to unlock progress in society, then we need humans to spend less time, you know, sending out faxes or filling in cells in Excel spreadsheets and spending more time actually thinking what is the problem?

How can we solve it? What are the right techniques for that? And that's something that even with the huge progress of AI, we're still going to need humans to do for a very long time. And then education and work, I think it's critical to re-educate people from the societal aspect, from the technological perspective. A lot of people are feeling left behind by the digital revolution. And how can we use AI to solve that? That's one of the key questions we want to ask and we want to challenge ourselves with.

And finally prevention and mitigation. That's where we're looking at predicting forest fires or reducing the probability of forest fires

which are becoming the worst thing due to climate change, droughts and so on. We can analyze large tracts of land and we can do statistical analysis to inform policymakers on how to better manage the climate. Love that. So key. Of course, we've all seen these devastating pictures of these fires blazing through Australia, through California in recent years.

My home and Paris's as well is Singapore and we've had horrible haze from fires that have been caused in Indonesia. Tell us a bit more about that there as well, Thiago. Could you share a bit more detail on how technology can help in preventing deforestation in general, forest fires or other sustainable moves that could help the forest and trees in general? Yeah, and as Portuguese, I mean Portugal is a country that's also heavily suffering from forest fires. So, there's something that's also dear to my heart.

But yeah, so actually recently we've been involved in a project with the support of Google Cloud, actually, together with our partners, IHI and Sumitomo Forestry, on managing peatland. So peatland is a specific type of forest where the soil is very...

wet, right? So it's almost like mud, right? And so it needs to, for the trees to grow healthily, the soil needs to be kept very, very wet, very irrigated, right? And it can be challenging sometimes, especially when the forest is being also used for commercial. That's peat, right? P-E-A-T, you mean, yeah? Yes. Peat, yeah, got it. I'm with you. Yeah, carry on.

So when the forest is being used for commercial exploration, sometimes it can be hard because the ecosystem is being changed. And the question is, how do you manage that? Right. And our clients, they've installed irrigation canals. Right. But those then they span a very large tract of land and it's very hard to manage manually. So they have staff on the ground, but...

What they end up doing is just going around and then opening one canal and then closing it. And it's mostly based on intuition. And sometimes intuition is not enough, right? So some areas of land can end up being too dry. It's like hope is not a strategy and neither is intuition, right? Exactly, exactly. I mean, considering...

you know, what they are doing. They are actually doing an exceptionally great job, but it's just that the area to be managed can get too large, right? And the companies we're working with are actually considering extending this strategy to bigger and bigger plots of land where, to be honest, right now, there's just not enough people to manage them, right? And so even if this plot of land in particular has someone to look after it, a lot of them don't. So the hope is that we can actually responsibly manage very big areas of forest

And so, yeah, like where we came in as AI providers is to help them with this management of the irrigation. So what kind of irrigation patterns do you need to minimize the likelihood of forest fire? In other words, how do you water the forest? I mean, in the very concrete terms, how do you water the forest optimally? Interesting.

And to do that, we're developing one of these, as I mentioned before, neural network simulators. So you can, based on the data and the knowledge of the physics of the water diffusion and the dynamics of evapotranspiration of trees, you can create a simulator of, okay, if I have water flowing through these canals and if I don't have water flowing through these canals, this is how the soil is going to look like.

And based on that, then we can make decisions. Did you say evapotransportation of trees? What was that term? Evapotranspiration. So basically the trees are sweating. Oh, nice. I prefer that term. Hey, Ferris, I don't sweat. I evapotranspirate. That sounds way nicer. Great term.

So, Tiago, that is the project that you work with Sumitomo and AHI, correct? Yeah. So when you first start, you mentioned about the problem that they were facing. You came in as an AI partner. So when you first start the project with them, working on all of this fantastic solution that you have, any challenges in particular that you have faced and how did you address them? Well, one of the main challenges was the data collection. And I feel like that's a challenge with most AI projects.

companies tend to not be prepared for AI. So they tend to have areas that are obviously AI and areas that are obviously not AI. And forestry tends to be obviously not AI, right? So they weren't expecting that this is an area that would be the target of AI modeling. And that means that the data is scattered. It's manually, there's hand-drawn diagrams, there's files scattered throughout, right?

And a lot of it in the beginning was working with the clients to help them systematize the data, understand what is available, what needs to be collected. And then given the availability of data, because the real world is not ideal. So, you know, as an AI engineer, I don't get all the data I would like. Of course, I would like them to install sensors everywhere and have drones flying over every second to collect data. But unfortunately, that's not economical or practical, right? So we also need to adjust how we're doing the modeling

to the amount of data that we have. And so working with them, and it's been a very enjoyable partnership because they're very eager and willing. They sent us photos from the terrain and got even an expert from Indonesia to fly over to Japan and talk to us. So it was very easy for us to acquire the domain knowledge that was necessary to systematize all the data and then start developing the AI model.

And just switching a little bit into the long-term perspective there as well, Tiago, obviously sustainability is something we all want to take place in the long run as we, you know, I think 2030 is a big marker and all the other initiatives that are happening globally. But from your perspective, what are some of the measures perhaps that you can take or have taken to ensure solutions for sustainability that can last?

Right. So as part of our mission, we really care about sustainability and we work with a variety of partners and we're developing AI solutions for all of them. I think one thing that's very important to us is to make sure that those solutions live up to certain standards that we check in advance. And we always like to check first whether the solution is having a positive impact on sustainability.

Every stakeholder, right? So not just our customer, but what about our customers, users? What about third parties? And key to that has been establishing an open communication within the company and allowing every employee that is involved with the project to voice their opinions, right? So it's not just that the management decides we're going to do this project.

But in reality, we tend to share the thought of the project, the overview, the ROI for our customers, what we're going to do. And we let engineers, we let salespeople, everyone pitches in and suggests, does this meet our standard for what we should do or not? Are there ethics there? Is the decarbonization there? So for example, for one of the projects we've done, we did a two months study with their data where we actually verified with implementing our solution,

actually cut carbon emissions and we have shown how many carbon emissions would be cut and based on that then we made the decision to go ahead right so i think a lot of thought needs to be put into projects and this is not something just for a company like us that's sustainability first but i feel like every company that is implementing ai solutions can do the same which is to do some kind of impact study in advance you know how when we do the large infrastructure projects

We do the ecological impact study, but for AI, we should have the same, right? What is the impact of this, both from an economic perspective, which of course business owners tend to be very focused on, but also from the sustainability perspective, the impact on all the stakeholders. Love that. Thank you.

Yeah, I think that's fantastic. And since the beginning of this conversation, you're sharing from you wanted not just to do a job, but a job that make great impact. And then you choose into sustainability. And now, you know, we have forest and sustainability and then AI. So all of these pillars, when they come together, education play a very important role.

So do you agree? And how do you think technology can help to achieve this? Right. I think there's a lot of consideration. So as you guys are familiar right now, there's a big hype with the large language models and being able to automate a lot of work that previously was done manually by humans. And so I think nowadays it's becoming even more critical to understand how do we as humans stay relevant? A lot of people have come to me and asked me,

are we even relevant anymore? Do humans matter? And I do believe that the answer is yes. But at the same time, I believe that we cannot just keep doing the same thing that we've been doing for the last 2000 years, right? So the role of humans will keep changing. And critical to that is that principle that

We need humans to propose problems to solve and to be creative about how to solve them, right? So it's not just about the implementation. So the implementation, we can probably leave most of it to AI, but it's about what is the challenge, right?

What is the new idea that we're going to try? What is the hypothesis? And how do we test that in the real world? And how do we measure that impact? So we need humans to do that, right? And at this point, maybe a lot of people still don't understand technology to the point where they are able to meaningfully contribute to that.

And that doesn't mean because they're not skilled enough or something. It's just because they've had other priorities in the past, right? They've had to make their ends meet. They focused on other areas. But now some of those other areas are becoming less relevant and technology is becoming more relevant. So I think...

That's where we need to step up as an industry, not just recursive or whatever. But I think big companies, I think Google is doing a great job and other companies, Microsoft, Apple and so on, doing a great job at trying to support communities to educate themselves at the technological level. But beyond that,

It's not just about learning programming, but it's about learning a mindset, right? Learning a learning mindset, kind of. It's like a meta concept, but learning to learn. I think that's the crucial thing that is going to keep humans relevant. And that's where I think we want to focus on. And we want to develop technologies with our partners that can stimulate people to develop that mindset through the use of, you know, large language models or other techniques where we can create

engaging and stimulating curricula that people can follow and they can use to pick up what is technology, what is sustainability, how to innovate and how to learn. A mindset or even a deep mindset as Google will have us believe. Right.

But you're so right there. That's such a meta skill, right, Tiago? Learning to learn or the skill of learning is one that people don't often talk about. My old boss used to say, if you're green, you're growing. And if you're ripe, you're rotten or you're rotting. And it's like, I think that very much speaks to that.

Time is absolutely flying. I knew it would because there's so much that we could talk about. And I really am enjoying this intersection that you've discovered with recursive AI on obviously sustainability with AI. As far as the future, though, if we were to kind of close with this one, like if you could wave a magic wand, where would you see AI in the big picture? And how would you see the world, you know, five, 10 years from now, if you had your way?

Well, I think it fits into this vision of creating a better future on the basis of technology. So the way I would like things to develop is that we are able to use these technologies productively to solve problems. It is a fairytale wish because any technology can be used for whatever purpose

the person using that technology can be used. And, you know, humans have varying amounts of purposes, right? But if we have a fundamental faith in humanity that most people tend to want to make things better, then having more powerful technology is in net a good thing, right? And I believe that with the explosion of AI for natural language processing that can learn to automate mundane tasks, I do honestly believe that

people will have more time for creativity and I believe that even though people sometimes think oh an idle person is is a wasted person but in reality people do like to be idle for a while but nobody likes to be idle for too long right and if you let someone bored long enough they're going to start to come up with creative new solutions they're going to want to make their communities better they're going to want to create art they're going to want to create science right and so I think

My ideal for the next five to ten years is that AI is able to release us from those kinds of mundane tasks and we're going to be able to use that technology productively to kind of step up our society and start to come up with scientific, artistic and engineering solutions to the problems that we really want to solve both at the local and global level.

Love that. And why not, if it's all making the world a better place and taking care of our planet, which you are very much doing. Thank you so much, Tiago, for joining us and for sharing all the great stuff that's going on with recursive AI. And just love how you're merging these two amazing pieces of importance to the human race in this day and age and into the future.

And this was all recorded by real humans, right? We didn't have AI write the script. We didn't generate the questions or answers so the audience could feel good with that. But great to have you on the show. Yeah, thanks for sharing, Tiago. And all the best to your journey in both AI and Japanese learning. Thank you so much, Dion, in Paris. It was a pleasure to have you. Brilliant. Take care.

And thanks to all of you who are listening, that digital show, APEC, will be continuing to deliver unique and engaging content. And talking about that, what is the one part that we cannot miss, Theo? Oh, wow. Well, so much there, Paris. I mean, I loved what he said about AI removing humans from the shackles of mundane tasks to allow for creativity in art, in science, in who knows what.

And that really is, I love that. It's a glass half full phrase and it really shows you there's absolute truth behind that. It may take time, but humans have been fearing AI since the dawn of, not say AI, but automation. Ever since the factory was built, people worried about their jobs. Ever since machines were built, we're worried about the job and not AI. And here we go, right? Yeah, but I really enjoy the part where he mentioned, you know,

People need machine learning, needs AI, but at the same time, AI needs humans as well to point out where's the problems and where to fix. There it is. The cooperation and collaboration of the best that our world and our future can bring. And it's happening now.

Well, with that, Paris, I think we need to let the audience go. But before that, of course, we should at least taunt everyone with a bad joke. And I have one up my sleeve. I cannot wait. All right. So I actually hear that they've made a new artificially intelligent Oreo. Do you know what it's called? Hit me. One Smart Cookie.

You like? No? Oh, you're being nice to me. I don't think that was very good. All right. If you have... It's hit the point of bad joke, but yeah. If you have a bad joke, question, comment, or anything else, please hit us up on social media and join us next episode as we explore another thought leadership topic on That Digital Show. Bye. Bye.