Season 6 Episode 8 Dec 23, 2021

The Web3 Debate and the World’s First Organic Reproducing Robots


Getting caught up in the web3.


In this episode, we talk about some criticisms of web3, which Tesla CEO Elon Musk, and Twitter co-founder Jack Dorsey wrote this week, and then we’ll speak to Samina Kabir, developer advocate at Decentology and Niharika Singh, product manager at Decentology, a web3 company about what in the world web3 is and their perspectives on it. Then we’ll speak with Sam Kriegman, postdoctoral fellow at Harvard and Tufts about his research with the world’s first reproducing organic robots.


Saron Yitbarek

Disco - Founder

Saron Yitbarek is the founder of Disco, host of the CodeNewbie podcast, and co-host of the base.cs podcast.

Josh Puetz

Forem - Principal Engineer

Josh Puetz is Principal Software Engineer at Forem.


Samina Kabir

Decentology - Developer Advocate

Samina is a Developer Advocate at Decentology on a mission to make it easier for beginner developers to learn about and build in Web3 through the Hyperverse. She is passionate about Decentralized Finance and DAOs and how they will increase equity across the globe. Samina contributes to organizations such as Developer DAO, TryCrypto, and CodeClubs in the intersections of community building and education.

Niharika Singh

Decentology - Product Manager

Niharika Singh is a passionate web3 technologist, an avid learner, and a community educator. Singh has been active in the blockchain industry since 2017. Currently, as a product manager at Decentology, her focus is on shaping the future of Hyperverse, a platform that helps web2 devs transition into web3 seamlessly. In addition to her role at Decentology, Singh is also an active contributor to DAOs such as TryCrypto Collective and ScribeDAO.

Sam Kriegman

The Allen Discovery Center at Tufts University and the Wyss Institute at Harvard University - Postdoctoral Fellow

Sam Kriegman is a computer scientist using AI to create new kinds of robots as well as engineered organisms called xenobots.

Show Notes

Audio file size





[00:00:10] SY: Welcome to DevNews, the news show for developers by developers, where we cover the latest in the world of tech. I’m Saron Yitbarek, Founder of Disco.


[00:00:19] JP: And I’m Josh Puetz, Principal Engineer at Forem.


[00:00:22] SY: This week, we’re talking about some criticisms of Web3, which Tesla CEO, Elon Musk, and Twitter Cofounder, Jack Dorsey, wrote this week. And then we’ll speak to Samina Kabir, Developer Advocate at Decentology, and Niharika Singh, Product Manager at Decentology, a Web3 company, about what in the world Web3 is and their perspectives on it.


[00:00:43] SK: We’re adding these identities through like public, private keys. We’re promoting this idea of monetization and owning your own content. So we’re trying to move away from these monolithic entities that we saw come with Web2.


[00:00:55] JP: Then we’ll speak with Sam Kriegman, Postdoctoral Fellow at Harvard and Tufts, about his research with the world’s first reproducing organic robots.


[00:01:03] SA: They’re not something that we were trying to do. They’re not something that we wrote in a grant. We’re going to make robots out of frog cells. Please give us a million dollars.


[00:01:14] SY: If I had to give an award for biggest tech buzzword of the year, it would either have to go to NFT or the fairly new and nebulous concept of Web3, which has to do with a blockchain-based decentralized web. And this week, it’s gained even more attention due to the criticisms of tech billionaires, Elon Musk and Jack Dorsey. So on Sunday, Elon Musk tweeted out, “I’m not suggesting Web3 is real. Seems more like a marketing buzzword than reality right now. Just wondering what the future will be like in 10, 20 or 30 years. 2051 sounds crazy futuristic!” And then the next day, Dorsey shot out a tweet saying, “You don’t own Web3. The VCs and their LPs do. It will never escape their incentives. It’s ultimately a centralized entity with a different label. Know what you’re getting into.” So obviously both Musk and Dorsey don’t believe that there can be a truly decentralized web that isn’t owned and controlled by a handful of the rich, at least the way Web3 is talking about it. But it’s not just tech CEOs who have their gripes about the concept of Web3. There was an interesting piece in TechCrunch just last week titled “The irrational exuberance of Web3,” where contributor Ron Miller goes into detail about the hype around the term Web3 and how it could just be a repackaging of old blockchain technologies and ideas that have been around for a long time, but in a way that sheds some of their negative associations. Coming up next, we speak with two employees of a Web3 company to dive deeper into the concept of Web3 and to get their take on it after this.




[00:03:08] SY: Here with us is Samina Kabir, Developer Advocate at Decentology, and Niharika Singh, Product Manager at Decentology. Thank you both so much for joining us.


[00:03:17] SK: Thank you for having me.


[00:03:18] NS: Thank you.


[00:03:19] SY: So tell us about Decentology.


[00:03:22] SK: Awesome. So we’re working on building a marketplace for smart modules or smart contracts to help developers be able to build on Web3 a lot easier.


[00:03:32] NS: Samina put it right beautifully out there. So our mission at Decentology is to onboard the next 10 million Web2 developers into Web3. And the way to do that is that we created this product called Hyperverse. And as Samina said that it’s an open marketplace for composable smart modules. So essentially, you can think about it like if you want to use a service, you pull in that smart module/smart contract into your app and you can just get it going with it.


[00:04:00] JP: For those of our audience and frankly me that don’t quite understand what Web3 is, could you define it for us? And you mentioned Web2. Let’s start with Web2 and then move onto Web3.


[00:04:12] SK: Actually, I want to go back to Web1 even.


[00:04:14] JP: Oh, let’s do it. Yes.


[00:04:16] SK: Awesome. So like in the beginning of the internet, we had this basically read-only static system where you would go on the internet, a couple of developers would upload something and all you would do is look at a webpage or reference something, any sort of data that was on the web was from a static file system. And so we added on another layer called Web2 where you started adding interactivity. So now instead of just reading from the web, you are able to read and write. So this brought upon different applications like social media, blog posts, forums, things like that. So now you’re interacting with the web. One of the problems that it brought out is like over time, we saw that you need to have hosting services and other things. So it started to become more centralized and monolithic. So several entities kind of like are pretty much like siphoning the servers and all these things. So what a Web3 does, it comes in with another layer where we’re adding these identities through like public, private keys. We’re promoting this idea of monetization and owning your own content. So we’re trying to move away from these monolithic entities that we saw come with Web2. And that’s kind of like my simple, non-technical explanation of it.


[00:05:28] SY: So you mentioned Web1 read-only, Web2, now I can comment on that blog post and I can do social media. What does Web3 look like? I get kind of on the backend this idea of centralization versus decentralization. But in terms of what the user can do, in terms of what we experience as citizens of the internet, what is going to be the difference for us?


[00:05:52] NS: So the key difference between Web 2.0 and Web 3.0 actually lies in the technological architecture of it. As a user, at the end of the funnel, you might not see a lot of changes in the future. Right now, it’s a bit chaotic, but eventually when user interfaces would improve, it would most likely stay the same. So from the technical point of view, the main thing in Web 3.0 is that there are no centralized servers, databases, clouds, DNSs or content delivery networks, CDNs. So essentially they would all be decentralized. So the community that’s taking part in so-called protocols by running their own nodes/computers, they are essentially running a program and there is no one computer or no master computer or master node that exists. So in that 3.0, it’s all distributed in a very diverse manner. So if I just give you a very small example there. So when a user tries to hit a URL in a web browser, the request is first and foremost broadcasted to a DNS and then the function of the DNS would probably be to resolve it to a URL and an IP address. And this IP address would in turn point to some server on the web. And that’s where the application logic of that website really lives. And in case of a 3.0, that DNS or the application logic would not only be residing on one master node, but essentially on multiple nodes or peers and that’s from where the data would be coming from. So the TLDR would be that there’s no one master node, but rather many, and that’s the key difference from Web2 and Web3.


[00:07:38] JP: Why is decentralization an advantage for Web3? Why is that important to end users and to app developers?


[00:07:47] NS: If I talk about the end-users, Web3 gives you more privacy. It’s much easier to operate anonymously behind an address, a blockchain address. In Web 2.0, that’s practically impossible. You have to have some kind of a Google or Facebook account, and then you have to log in through that or create like a whole different thing. Secondly, in Web 2.0, when I said that there’s like a centralized node, it means that there’s degraded security. So you only have to attack one node to compromise for the security. In Web3, it’s all decentralized. So the cost of attacking a network would be much higher. Another thing that comes to my mind is controversial usage of data in Web 2.0, like companies have been exploiting users’ data, selling it and doing whatnot with it. In Web3, that will probably come to a stop. That being said, censorship in Web 2.0, like some person, some companies, some authorities sitting there who were monitoring your data and as long as you’re churning out data according to their policy, well and good. And the second you start speaking your own mind, so to speak, they will put a ban on you. So there’s a lot of centralization of power that we see and that’s pretty harmful/toxic in the longer run. So these are some of the disadvantages of Web 2.0 so to speak.


[00:09:10] SY: So I know that when a lot of us, including myself, think about Web3, we think about cryptocurrency. Can you talk a little bit about the relationship between Web3 and cryptocurrency, but it also sounds like there’s more to it than that? So can you speak to things about Web3 that are not related to crypto at all?


[00:09:28] SK: So originally, we had Bitcoin, which was just cryptocurrency. It was an application of the blockchain just doing that. And then later on and around I believe it was like 2015, 2016, I might be wrong on the days there, we added Ethereum, which was similar to Bitcoin, but now you’re adding the ability to build like software or applications using this blockchain technology. So we saw a lot of use cases come out of it. One of the primary ones that we see a lot, especially on media and if you go almost anywhere these days is NFTs, which are Non-Fungible Tokens. And so these are basically a way of verifying ownership of some metadata. So this is a use case that I think blockchain has that’s not directly related to cryptocurrency. Like you can use cryptocurrency to buy the NFT, but the NFT is also started on blockchain and the use cases can vary from things like social tokens, selling artwork, just anything that requires ownership. I think in the future we’ll see use cases like real estate. Maybe like businesses will do all the work through like NFTs or smart contracts. So this is that other additional layer that’s not cryptocurrency, but that’s also done on the blockchain.


[00:10:39] JP: So what about Web3 excites each of you the most? Either the technologies in Web3 or the applications that are enabled by it do you find most compelling?


[00:10:51] SK: The first thing that really compels me is DeFi. DeFi is decentralized finance, basically relating to cryptocurrency. With decentralized finance, one of the things that really excites me is globally we have 1.7 billion people unbanked, where they don’t have access to banking and banking is really, really important, not only for investing, but storing your funds, being able to move funds and do things like that. And this is global, right? This is a global problem. So decentralized finance is a way where you don’t have to like, the traditional banking system, where you have to go in with your name and your assets and you have to like store everything and give everything to the banks, you can create a wallet and you can start participating in DeFi anonymously and also be able to do all of these things that you do in traditional banking, but do it through DeFi and it’s way more accessible. So that’s one application that really excites me. The other one is DAOs, which is Decentralized Autonomous Organizations. And these are really cool because I like to think of it as like a more democratic business where like everybody has the ability to participate through their tokens. They’re able to vote on legislature. They’re able to participate and have ownership of whatever goes on instead of having like a centralized board that kind of makes decisions for everyone. Everybody has that ability to participate. So that’s another application that really excites me. I think for traditional structures that we have now, like NGOs, like DAOs will be very, very useful in the future.


[00:12:17] SY: And Niharika, what about you?


[00:12:18] NS: So for me, the most exciting part about Web3 is that I think that it’s not just the technology movement. It’s a shift in culture. It’s going to change the way we interact with the web. And the aha moment, that moment that I had with a Web3 was also in decentralized finance. The exact reason why it was that as someone, imagine you’re in let’s say in India and you have your laptop and your internet connection and you just need a loan, let’s say for something, and then you go on a certain protocol. For example, you go to Compound and Compound doesn’t want to know your first name, your last name, your correspondence, address, your permanent residential address, how old are you, what do you do, what’s your current score. Without all of this information, you can still take out a loan. And that’s truly banking the unbanked people. In theory, it sounds like, “Oh, wow!” But when you actually do it in practice, it just feels fantastic. And that was like a wild moment for me. And that’s when I realized that this is not just the technology. It’s going to change people’s lives. It’s like going to be a cultural shift. And since then, everything I stumbled upon in DeFi, it’s like tying back to all the cultural things. And eventually, it doesn’t matter where you live in the world as long as you have an internet connection and a laptop, you can have access to the best of things because blockchain/Web3 doesn’t discriminate.


[00:13:53] SY: So Web3 has a pretty big promise built into it that blockchain technologies are supposed to break us away from the dominance and influence of a handful of companies and billionaires and the influence they have on the web. What are some of the major ways these technologies are trying to fulfill that promise? Is it kind of simply in this decentralization idea or is it more than that?


[00:14:15] NS: I would say that definitely the power is taken away from the hands of the big billionaires and it’s being put back with the people. And I’m not saying that whatever the billionaires did was wrong for wherever they got, like they got there because of their own hard work and I appreciate all of that, but sometimes by exploiting somebody’s data, by selling it to another company, that almost becomes unethical. And a lot of universities and research groups across the world are talking about it, how to make AI fair, and this and that. It’s just a coffee time thing, but nobody’s actually doing anything about it. And decentralized movement or decentralization is actually doing some ground work. And time and again, we see politicians putting it down, especially Elizabeth Warren, the US senator, sometime back like a couple of days ago, she said that, “Oh, DeFi is a scam and this and that.” I think that’s the fear these people have that wherever the power is centralized, they’re fearful that when the world will turn into a bankless world, they will be out of jobs. And of course, they have all the right to say so, but I think it is because of decentralization that all these authorities and big people with all the money and power, they are kind of looking at it now and it’s not looking good for them.


[00:15:44] JP: Let’s talk a little bit about Decentology. Decentology’s mission says that it’s to empower developers to build awesome decentralized apps. Can you tell us how Decentology does that and what do those apps look like?


[00:15:58] SK: So a decentralized application is basically like when we think of traditional front-end and back-end applications or a full stack application, we have our HTML, CSS, JavaScript, maybe some JavaScript framework like React, and then on the back end, we have all our database with Node and SQL and for like storing all of these different like points. But when we talk about a decentralized application, it’s just a front end where you can interact with the application. Most of the time you have some sort of wallet that allows you to authenticate and log in or use the application, but the back end is the blockchain. So it’s kind of different in that regard. Decentology is making it easy to build these decentralized applications. One of the pain points that we see is when people enter in the Web3 space is, well, one, it’s kind of a shift. So like when you’re talking about the back end, of course, but also it’s a little bit difficult to build because we don’t have these APIs and all these little different building blocks established yet. So I think that’s what Decentology is trying to do is build these little Lego blocks where developers can pick and pull different smart modules or smart contracts into their application and be able to build that easily. Another important aspect of Hyperverse, what we’re building is that these smart contracts will be audited. So we want to make sure that you are getting code that is not buggy. It doesn’t have an exploit. We want to ensure that developers are getting that quality experience when they are building their applications, if Niharika wants to expand on that a little bit.


[00:17:26] NS: I think you put it beautifully, Samina. I would reemphasize the features of Hyperverse that, yes, in the ecosystem right now, there doesn’t exist any project that’s catering to smart contract and possibility. With DeFi, you're seeing the term, “Money Lego”, being very popularized, but it’s just for finance and it’s not truly Lego until and unless it’s on the code level. With Hyperverse, we’re actually trying to solve it at the code level with smart contract and possibility. The next thing that Samina pointed out was the monetization part of it. So if you, as a developer, create some smart module, which is used by other developers in their applications, you can make some money out of it each time a transaction is called through your smart module. So it can get an economy going. The next thing is that we also have a marketplace where you can discover different smart modules based on your needs. If you want to have something for payment, you search for payment and a bunch of smart modules come up. And if you want to search for something else, it would be similar to what NPM registry is. Finally, the fourth feature that’s there is the auditing as Samina mentioned, you as a developer doesn’t get on block code. And the way that works is that people stake their Hyperverse tokens and each time there’s a bug detected with this contract, those tokens get slashed. The framework around this is still a work in progress, but overall, these are the benefits of Hyperverse as for you as a developer.


[00:18:59] SY: So what does open source look like in a decentralized Web3 world? Are they compatible?


[00:19:05] NS: Yes. That’s the case as of now. Like for example, some libraries such as OpenZeppelin, super famous for Ethereum development. They write code for various different standards. For example, if you want to create an NFT app, you refer to ERC-721 and that’s curated by OpenZeppelin. So yes, that phenomenon of open source in Web3 already exists, but it’s very similar to the open source in Web2.


[00:19:32] JP: Got you.


[00:19:32] NS: No innovation there per se.


[00:19:35] JP: What do you think about some of the tweets this week that were put out by some of these tech billionaires we’ve been talking about, in particular Tesla CEO, Elon Musk, and Twitter Cofounder, Jack Dorsey? And I think this gets back to our conversation. You were mentioning, we were asking about like Web 2.0 has been dominated by large companies and large tech CEOs and billionaires. And I guess a question I have is I don’t expect them to go quietly. Right? You don’t expect Mark Zuckerberg is going to be like, “You know Web3, that’s great, decentralized, and I don’t get any money. That’s fantastic.” There’s going to be some pushbacks. So I guess is the Web3 industry seeing that pushback? What is that looking like? Is it legislation? Is it technology? Where are the barriers from the existing entrenched Web2 people?


[00:20:23] NS: I mean, yes, we’ll definitely get pushback because Web3 is so young, like barely like half a percent or maybe less than that of the world population is working in Web3. You can like put them in one place and that wouldn’t even make a proper country. I would say that in the beginning, a technology is usually always ridiculed. Remember when the first cars came out, it was still like in the bonnet. You would still have a horse. So it would just look like a car, but the bonnet would not have an engine. It would still have a horse. What’s happening right now is that any press is not bad press. So even when Elon, like Elon, I don’t understand. He’s like always in his two minds, like first he’s promoting Doge and Floki and now he’s like, “Has anyone seen Web3? I can’t find it.” And Jack, he’s also saying such things, but I think this year, 2021 has been very good in its Web3 adoption. This was the year of NFTs and partially the reason why Facebook changed its name to Meta. I’m not saying that’s the only reason, but partially why? I think so. And more and more people are getting into it. The rise in it has been like, I don’t know, I read somewhere, it has been upwards of a hundred thousand percent. That’s not nothing. That’s probably something. Right? And no matter how much people say Web3 is dumb and stupid, everybody’s still reading about it and it will probably have some future. It will not die down that easily.


[00:21:53] SY: So there has been also criticism that Web3 is really just a marketing term, kind of repackaging blockchain technologies, like just a different way of saying blockchain technologies, but having less of a negative connotation, but it’s not quite a different concept altogether. What is your response to that? What do you think about that?


[00:22:11] NS: I think that’s true because I heard about Web3 only when blockchain came out. And that’s only when Web2 was coined and Web1 was coined. Before that, I had never heard about these terms. I mean, from a chronological point of view, I think it’s a great way to put it like Web1, Web2, Web3. It shows a great continuous upward innovation through it. And yes, Web3 is mostly about blockchain as of now, but probably in the future, it will probably also incorporate technologies such as AI, AI with blockchain or internet of things with blockchain. So I’m not sure how it’s going to evolve, but it’s going to be certainly more than blockchain.


[00:22:51] JP: Is there anything else we haven’t talked about with Web3 yet that you’d like to cover?


[00:22:56] SK: So I do get a lot of questions on Twitter ranging from like, “What is Web3? How do I start building in Web3?” And that’s something I like to address a lot. So if you want to build in Web3, there’s a variety of resources out there. One of the ones I really like is the University of California, they have a blockchain fundamentals course, and this is where I started learning about everything. So I’m going to give a shout out to them. I would definitely check out their course. But yeah, learning blockchain development, it’s not going to be a complete overhaul of what you’ve learned. So if you’re a developer and you’re looking to get into Web3, it’s just building onto your stack. So you come in with your front-end applications, you come in with your front-end stack of HTML, CSS, JavaScript, React, Spelt, whatever it is, you can come in and then pick up a blockchain language like Solidity or Cadence or Rust. And then there you go. You can work your way towards becoming adapt developer. So there’s lots of opportunities here. We’re looking for people to build. I think that in the future we’ll see more blockchain developers. Right now, the percentage is actually, as Niharika mentioned earlier, less than 0.05% are full-time blockchain developers. So there’s definitely a lot of opportunity. A lot of people are pouring oil into the idea of Web3. So I would definitely explore it as an avenue for a career.


[00:24:10] SY: Wonderful. Well, thank you both so much for being here.


[00:24:12] SK: Thank you.


[00:24:13] NS: Thank you for having us.


[00:24:21] SY: Coming up next, we talk about some exciting new organic reproducing robots after this.




[00:24:41] SY: Here with us is Sam Kriegman, Postdoctoral Fellow at Harvard and Tufts, and Lead Author of the research paper titled, “Kinematic self-replication in reconfigurable organisms.” Thank you so much for being here.


[00:24:54] SA: Hi. Thanks for having me on the show.


[00:24:56] JP: So tell us what your paper is all about.


[00:24:58] SA: Okay. So you mentioned this paper and it has a long and difficult to remember and reiterate title. And this was the third paper in a series where we were exploring these devices, for lack of a better term, that are somewhere in between robots and organisms we’re not exactly sure what to call them yet, but they’ve been popularly dubbed “Xenobots” and there’s no turning back. So as you mentioned, we call them reconfigurable organisms in the paper.


[00:25:32] SY: That’s not as much fun.


[00:25:33] SA: That’s not fun, but it’s formal and scientists like that because it describes basically that they are a bunch of parts from an organism, specifically one-day old frog eggs, and we reconfigure those frog cells into things that aren’t frogs, but can move and do some interesting things. The frog is called Xenopus Laevis. So they’ve been called now Xenobots.


[00:25:58] SY: So where did you get the inspiration to pursue this type of research?


[00:26:03] SA: So the Xenobots are interesting because they were a surprise. They’re not something that we were trying to do. They’re not something that we wrote in a grant. “We’re going to make robots out of frog cells. Please give us a million dollars.” I’m not sure anyone would fund that actually.


[00:26:23] SY: I don’t know. That sounds pretty. I would fund that. That sounds pretty.


[00:26:25] SA: Well, thank you. I’ll send the next grant proposal to you.


[00:26:28] SY: Deal.


[00:26:30] SA: We were trying to do bio-inspired robotics, which has been done since the very start of robotics in the late 1940s, which is basically you see some amazing animal or plant in nature and then you try to build a robot or some kind of machine that copies aspects of it. And this project was funded by DARPA, which is the research arm of the defense agency here in the United States. And I don’t know what DARPA exactly is looking to get out of it. They fund all kinds of things that have nothing to do with the military. They fund breast cancer research. They fund therapies for radiation poisoning and things. So they do a lot of really good things. And I can tell you just to try to head off some of this, the Xenobots are not weapons. Basically we would be doing this research no matter who was funding us. It just so happens to be DARPA.


[00:27:27] JP: So tell us, how exactly do you create these living robots? You said that they’re made from frog eggs.


[00:27:35] SA: Yeah.


[00:27:37] JP: What’s next?


[00:27:39] SA: That’s it. Don’t you get it, Josh? That’s it.


[00:27:42] JP: I mean I grew up next to a pond and there were frog eggs and I feel like I'm missing the stuff.


[00:27:47] SY: You missed your opportunity, Josh.


[00:27:48] JP: I did. I totally did.


[00:27:49] SY: You could have totally done this. You missed it.


[00:27:52] SA: Yeah. So you grew up next to a pond. You could make Xenobots if you wanted to. It’s really simple. It just takes a steady hand and you are essentially treating these frog cells like very small Lego bricks, and you’re looking under a microscope and you’re pushing them together and scraping them and burning them and doing whatever you can to mold them into different kinds of shapes. And depending on the shape that you make out of this frog tissue, the Xenobots will exhibit different behaviors. Essentially you take one-day old frog eggs, this means that they were fertilized 24 hours ago. And you cut off a very small portion of the cells from that egg. These cells are called the animal cap because they sit at the top of the frog egg. And those cells, if you don’t do any kind of biological magic to them, you just leave them on their own, they will grow into frog skin cells and they will also have very small hairs on them called cilia. And these cilia basically are like flexible paddles. They beat back and forth and they can propel this ball of frog skin cells around the dish. But just remember that we’re talking about something super small, less than the size of a poppy seed covered in very small hairs, swimming around a dish. It’s very, very simple and it comes kind of naturally from just sticking together the frog’s skin cells. We can make them out of different kinds of motors and different kinds of materials, but this is the simplest version of a Xenobot.


[00:29:27] SY: Okay. So you have these skin cells and you’re putting them together, as you described, like Lego parts and building these shapes out of them, and they do different things depending on the shapes, which I definitely want to get into. But are there any machine components to these robots? Is there anything mechanical or man-made that you’re adding to this design? Or is it just straight cells?


[00:29:52] SA: These are a hundred percent organic, non-GMO frog cells. They’re artisanal little robots. We’re not adding any synthetic components like plastic or rubber, although you could, if there was a reason to do that. But what’s nice is that by making them entirely out of living tissues, you get a lot of nice properties for free, such as they’re biodegradable, which is amazing for a technology.


[00:30:24] SY: Yeah, that’s a good point. I love that. These are environmentally friendly robots. This is awesome.


[00:30:31] SA: They are. You just have to convince people they’re robots. So biologists might say, “Of course, they’re biodegradable. It’s just a bunch of frog cells.” So they are still engineered and designed to behave in very specific ways. And for many that’s what makes something a robot. And the way that we do that is we use an AI system. It’s just a very simple computer program that consists of a simulation that’s kind of like a video game and a system of trial and error. And we just try out many, many different shapes, donuts, pyramids, spheres, footballs, whatever, you name it. The computer has tried it. And depending on that shape and what kind of motor cells we use, the Xenobot will behave in different ways. It might be good at walking or swimming or carrying a pellet or working together with other kinds of shapes to build a pile out of debris in their dish. And this all depends on their shapes. So in a way, the shape of the Xenobot is its program.


[00:31:34] JP: Okay.


[00:31:34] SY: That is really interesting.


[00:31:36] JP: Oh my gosh! We have so many things to ask. Okay. I think you briefly touched on this, but why do you refer to them as robots when they are all organic materials?


[00:31:46] SA: Yeah. I refer to them as robots because that’s the easiest way for me to understand what they are personally and how to control them. I think you could equally look at this as like an engineered organism, although it’s very, very simple. Some are just skin cells. Some are skin and heart muscle cells where the heart acts as like very small pistons to contract and expand and push the Xenobot about across the bottom of its dish. So there’s mechanical ways of looking at these things, but they are extremely simple. They’re robots because they are designed and engineered. I think that’s something that’s important. For it to be a robot, we can train dogs to do tricks. In fact, we could probably train a frog to be more useful than these frog robots that we’re making. But the frog you can select for over many generations to create frogs that maybe have slightly longer toes or have slightly different anatomical features, but no matter how long and how much you select for, it’s going to be very difficult to create something that you wouldn’t say that’s a frog. But with these Xenobots, you can say, “I want something made of skin cells and heart cells. I want the heart cells to go down here, the skin cells to be up here, and I want it to be in the shape of a donut or the shape of a quadruped, something with four limbs. You can change the number of limbs. You can change the muscles, the motors, the sensors, and you can keep starting over from scratch, which is something it’s difficult to do if this was just an organism that we’re trying to domesticate and evolve. So I think, for me, it’s easier to consider these as robots that if designed correctly might be able to perform useful work for us.


[00:33:31] SY: I see. I see. So let’s dig into the designing element. You touched on this a little bit with bringing in AI and to figuring out what shape they should be in and what components. Can you talk a little bit more about where software comes in? Where this computer science, AI, machine learning, all those things, where does that fit into your project and in building these robots?


[00:33:51] SA: Yeah. There’s a couple different ways to think about this. We could talk about how do you get a living system to do what you want it to do. Living systems have all kinds of great features like biocompatibility and they are self-renewing and self-powered. It’s very difficult to get a very small robot, the size of the Xenobots, which are smaller than a grain of sand. It’s difficult to get electronics and batteries in there. So basically, what we’re going to do is instead of plugging in a USB cord into a Xenobot and downloading some software into it to direct its motion, which is currently not possible because nothing’s in there except a bunch of a very simple cell types. Instead of the software, we’re going to design its shape and its distribution of these cell types. And you can get a lot out of that. You can create Xenobots to behave in very different ways that don’t need a brain yet. Maybe this will become a tool to realize when brains are really necessary. But at the moment, they’re not doing anything that requires a brain, which might be something like very complex decision-making and very fast rapid movements of their bodies in like surprising circumstances. But right now, they’re lazily swimming or walking around a Petri dish. So brain and software is not necessary. The software that we use is simply a design tool. It’s just a means to an end. So we want to get a Xenobot that behaves in a certain way. How do we do that? We could try out a thousand or million or billion designs, but building each one takes a long time. So we simulate them. We created a computer model that is accurate enough that it can kind of filter out the bad designs and find ones that are promising, but the computer predicts, will do what we want the system to do. And those are the ones we build. And after we build them, we might say, “Well, the computer thought that this Xenobot was going to jump up and down. That’s not what it’s doing. We need to adjust the computer model.” And we go back and forth a few times, but we’ve been able to create a simulation that is accurate enough that we can do a kind of AI design process inside of that simulation and then transfer the simulated designs to reality. There’s this whole field called simulation to reality transfer or sim-to-real with the number two in it. And roboticists have been doing this for a long time. So we’re just taking some of those tricks and applying them to biological materials instead of metal and plastics.


[00:36:36] JP: Are these cells capable of reproducing on their own or are they completely constructed right now?


[00:36:43] SA: So the cells cannot reproduce. And we didn’t think the Xenobots would be able to reproduce. They only have skin cells. They don’t have reproductive cells. So once we make a Xenobot with 5,000 cells, it’s in the shape of an ottoman or something and has four little limbs, for example. Some of the cells might die and some cells might divide a little bit. But overall, there’s going to be around 5,000 cells in that Xenobot for its lifetime, which is like about two weeks and then it starts to just biodegrade. We didn’t think therefore that the Xenobots could reproduce or replicate in any way because what happens in nature, depending on how you count, there’s anywhere from 2 to 9 or 11 ways that biologists have recorded in which animals and plants reproduce, and they all involve growth from a seed or an egg or a fragment of the parent and the parent donates this material. It has its genes in it or the parents donate this material. And from that rose an offspring and maybe there’s slight mutations in the genes and that’s how you get evolution. It wasn’t clear that Xenobots would be able to do this. We can cut a Xenobot in half, and there just becomes two smaller Xenobots, which is interesting in its own right.


[00:38:06] SY: Oh, interesting.


[00:38:08] SA: But they don’t grow to take the form of the original parent. There’s not really growth. Okay. So you have the system that doesn’t really have growth. How can you get it to reproduce? We turned to the robotics literature. There has been interest in machines building other machines since like 1948. This guy, a mathematician named John von Neumann, who, if you look at his Wikipedia page, what he’s known for is the longest list I’ve ever seen at anyone. It’s like every single thing that’s ever been invented in math and engineering, it feels like, he did a lot. And one of the things that he was just interested in, probably on a whim was machines that can build other copies of themselves. This was before the discovery of the structure of DNA. And it wasn’t clear that machines and animals were that similar. He thought that there might be some kind of like mathematical paradox of a machine building another machine. I guess the details aren’t that important, but he proved that it was possible for machines to build other machines. And because of that, this idea, these self-replicating machines are sometimes now referred to as Von Neumann machines. And we had the Xenobots and we noticed that when there was debris in their dish, they would move around the debris and they would push that debris into piles. So this was due just to the random movements. There’s no evidence that the Xenobots want to do this. They’re just moving kind of like windup toys or you can imagine Roombas that are moving randomly and there’s a bunch of Ping-Pong balls in the room or something. Maybe they’re a little bit sticky, so they don’t go flying everywhere. And if the Ping-Pong balls started out uniformly distributed in your apartment and you had a dozen Roombas moving around, I think it’s the statistical like mechanics problem that you’re going to start to get piles of Ping-Pong balls. The Xenobots, because of how they’re shaped and moving, they tended to make very large piles, which is interesting. And then my collaborator, Doug Blackiston, he had the idea of, well, instead of debris, they were silicone coated iron beads. It doesn’t really matter. It’s basically trash from the perspective of a Xenobot. Instead of that trash and the dish, you could put down other single loose frog’s stem cells and the stem cells are sticky. So as the Xenobots move around, these loose stem cells, they push them into piles of stem cells and those piles, if they’re sufficiently large, if they have at least 50 cells in the pile, it will compact into a sphere row, those little cilia hairs I mentioned that act as very small paddles on the surface, and then that small sphere of cells can move around. So that’s basically a child because now we take that sphere, we put it in a second dish, add more cells and it can push those cells in a pile.


[00:41:11] SY: Wow!


[00:41:11] SA: So it sounds like a party trick, but we hope that it kind of is enlightening to show that life can reproduce in another way. It’s definitely a proof of principle. It’s not useful yet, but we think that it’s more than just a trick. We think there’s serious science that we’ve uncovered here.


[00:41:28] SY: Tell me a little bit more about what these things look like and what they do, especially since you mentioned burning them and putting them into different shapes. You mentioned a futon, which is very interesting. What are some of the other shapes that you’ve put them in and how does that change what they do and how they operate?


[00:41:47] SA: One of the interesting things using AI to design systems, and again, I use AI. Some people might take issue with that. It’s very, very simple. It’s just that a process of trial and error, but it’s happening on the computer automatically. So this automated design process, when you apply this to physical systems or basically to design anything that humans normally design, sometimes the computer comes up with new unique creative designs that humans wouldn’t think about. So when we go to design robots, we design robots to look like familiar objects. And that’s why all the robots dancing on the internet look like dogs and humans. That’s what we see in the world. So that’s what we build, but the computer doesn’t share our biases. It might have its own bias, of course, but it’s at least different from a single human designer. So it can help diversify the design process, which is well-known to help creativity and improve this design process overall. So that’s one thing you get from using this computer system. The other thing that you get is sometimes frustrating. Sometimes the AI comes up with designs that are very simple. Humans tend to overthink things. So we designed very complex solutions to problems. The computer will come up with the simplest possible solution that works. And in the case of the self-replicating Xenobots, the computer came up with Pac-Man, which is just a disc with a little tiny mouth cut out of it. Super simple. So that ended up being a good shape for the self-replicating tasks because, as I described before, what essentially the Xenobots are doing is just moving around their dish, like small bulldozers and the Pac-Man shape is pretty good at collecting cells in its little mouth, moving them to other piles of cells and depositing them, and just basically acting as a shovel to create larger piles of cells, which correlate strongly with how much and how well, how strong that offspring will become out of that pile.


[00:44:03] JP: So I’m imagining because these Xenobots are just made of living cells, do they die eventually? What’s their lifespan look like?


[00:44:13] SA: In fresh water or slightly salty water, the Xenobots will live. They will move around and do whatever it is they’re designed to do for about 15 days or so, two weeks basically. And then they start growing old. They turn transparent and their shape starts becoming funky and then they start to slowly basically disintegrate into skin cells that are just dead, which is nice because all machines break down and fall apart. And if you are a machine made out of metal or plastic and you’re operating in the environment and you start to break down and deposit metal and plastic and chemicals, that’s not so great. But if we can get these Xenobots to do something useful in the future, we know at least they won’t add pollution to the environment as they’re performing their tasks, which is I think a good goal for roboticists overall.


[00:45:14] SY: So what are the potential benefits of having these living machines? I know right now it’s kind of a proof of concept and it’s super, super early, but where do you see the application for this? What do you see that being?


[00:45:29] SA: I think in the future, if there are any applications coming out of the Xenobots as we now understand them, it might be something like underwater applications. So I mentioned that machines made of plastic break down into smaller bits of plastic. These are known as microplastics and they’re found everywhere on earth. And there’s currently not a consensus on whether or not microplastics are bad for our health. Personally, I think that the default hypothesis should be they’re bad until proven otherwise, but that’s not the case.


[00:46:01] SY: I feel safer. Yeah.


[00:46:03] SA: Yeah. We’re saying, “Oh, well, we don’t know they’re bad. So let’s do a lot of tests to find out if they’re bad.” And by the time we finish all these tests, there’s microplastics everywhere. I mean, it’s already the case. So I hope that maybe robots could be used to clean up this mess that we’ve created. And I spoke about how Xenobots are good at aggregating debris, whether or not that’s other cells or essentially trash in their dish. Maybe they could be used to aggregate microplastics in waterways into a larger ball of plastic that’s easier to find and pick up and recycle by human or another kind of robot. So that’s one application that we can imagine, but it’s definitely science fiction at the moment because that would require a lot of Xenobots. The Xenobots are super small and to manufacture them on that scale would probably require automating their manufacturer. And at the moment, they’re all made by hand and 3D bioprinters don’t reach that resolution yet. So there’s work to be done. You also have to get a lot of frog eggs to make these. I don’t think that would be a problem, but there’s a lot of setting this up and convincing people to fund it. And at the moment, they wouldn’t be capable of doing that.


[00:47:18] SY: So are there any potential negative outcomes you can think of at this point? I know it’s super early days, so it might be hard to kind of think of those, but is there any downside to having these living machines?


[00:47:31] SA: I think that there could be many downsides to making robots out of living systems, if those living systems were suffering in any way, but these are taken from one-day old frog eggs, and there’s nothing in those eggs. It’s just a mush of blank slate, stem cells. There’s no neurons in there. So I feel like the moral calculus is good right now. That might change in the future. And we might have to decide that it’s not worth building these things. If the donating animal is suffering in any way, and as I mentioned right now, it’s just an egg with nothing in it, so I don’t think so. Or if the creature that we create has the ability to suffer, that would be something that we wouldn’t want to do obviously. Whatever this calculation is, though, we have to consider the other side of the equation, which is all the good that they might be able to do. So if it requires us to have this huge facility of breeding cells and there is some kind of ethical negative in the equation, but the outcome of this is that we can have really useful medical devices that save lives, we’re going to have to weigh the pros and the cons of this and just decide where it should be regulated, where it should be banned and what kinds of systems we want to have as a society.


[00:49:02] JP: Is there anything else we haven’t covered yet that you’d like to talk about?


[00:49:06] SA: We’ve talked a lot about biology, a lot about robotics, but there’s I think interesting stuff here on the coding side. So one is this example of virtual creatures, it’s called, and it’s essentially like a video game. These were first created by some people at MIT, Karl Sims and Jeff Ventrella, they were interested in computer graphics and it turns out one of the best ways to make a realistic looking character in a computer animated film or movie or whatever you’re trying to make is not to hand design every little step and movement inside of that character. But instead you just say, “Look, this is how physics works. If you have a ball on top of a hill, it rolls down. If you have a body and has this kind of mechanics and it has these forces applied to it, then it’s going to move like this.” So encoding the laws of physics into a computer system is a really useful way of getting lifelike movements. So that’s how it started, but didn’t take long for roboticists to realize that this could be a really useful way of designing robots that would then be built in the real world. So I think what’s interesting here is that this might not just be a useful way of designing robots. It might be a really good way of doing AI in general, because if you want to create an intelligence system, why not try to create it the same way that intelligence evolved on earth? So intelligence didn’t evolve on earth by sorting through trillions of images and labeling them or doing natural language processing. It evolved by systems just trying to adapt and survive in their environment. And one of the key things that we associate with intelligence happens to be brains, but brains didn’t evolve for determining whether or not it was a cat or a zebra and a picture. Brains evolved for movement. So if we can create systems that can move even in physical simulations really well and adapt to different, surprising circumstances they encounter in their physics simulation, then that little character that’s moving through there is going to embody some knowledge of the world that’s similar to ours and how we experience space and time. And once you do that, I think once you have down movement, the ability to deal with surprising circumstances as you’re moving through a world, the rest of intelligence I think is really easy. It took evolution really, really long time to get animals and plants capable of just surviving in the world. And not so much time after that, just adding on the extra stuff that makes us different from, for example, a frog that is a small, little step in evolutionary tree, took a lot longer to create cells and systems capable of just moving through the world. So I feel like maybe we should focus on that a little bit more than we have been.


[00:52:17] SY: Well, thank you so much for joining us.


[00:52:19] SA: Thank you for listening to my very long academic rants.


[00:52:22] SY: No. This was awesome. This was really great.


[00:52:25] SA: Thanks a lot.


[00:52:36] SY: Thank you for listening to DevNews. This show is produced and mixed by Levi Sharpe. Editorial oversight is provided by Peter Frank, Ben Halpern, and Jess Lee. Our theme music is by Dan Powell. If you have any questions or comments, dial into our Google Voice at +1 (929) 500-1513 or email us at [email protected] Please rate and subscribe to this show wherever you get your podcasts.