AI and Robotics: The Future of Intelligent Automation

Interactive Transcript

spk_0 Welcome back to Inside Assemble AI and welcome to our Season 1 finale.

spk_0 I'm Sam and alongside my co-host, Mack, we have spent this entire first season exploring

spk_0 how artificial intelligence is transforming business strategy and innovation across every

spk_0 sector imaginable.

spk_0 From AI in marketing and education to executive leadership and community building, we have cut

spk_0 through the hype to bring your real insights from leaders for implementing AI strategically

spk_0 in their organizations.

spk_0 Today, we are closing out our first session with what might be our most exciting topic yet,

spk_0 the conversations of AI and robotics and how this powerful combination is reshaping everything

spk_0 from disaster response to this exploration to the factory of floors and homes of tomorrow.

spk_0 We're asking the big questions, how is AI enabling robots to think, adapt and operate

spk_0 in ways we never thought possible?

spk_0 And what does this mean for our future, both the incredible opportunities and the challenges

spk_0 we need to navigate responsibly?

spk_0 Our Season finale episode is called AI and Robotics, the feature of intelligent automation.

spk_0 And we are diving deep into the technical breakthroughs, real-world applications and ethical

spk_0 considerations shaping this rapidly evolving field.

spk_0 Today, we are joined by Daniel Riching, a veteran technologist and founding member of Brain

spk_0 Wave Collective, where he explores the intersection of AI robotics and advanced emerging

spk_0 technologies with over 15 years of experience in DevOps, software integration and system

spk_0 design.

spk_0 Daniel brings a technical perspective grounded in hands-on, experimentation and performance.

spk_0 He is committed in church, numerous hackathons internationally, including events hosted by

spk_0 Huggings Fest, Nvidia and Meta, often earning top owners for novel applications of AI in

spk_0 hardware and software systems.

spk_0 His recent work includes contributions to open-source AI tools, networked robotics and

spk_0 agentic video frameworks.

spk_0 Daniel's experience spans both large-scale enterprise environments and fast-moving prototype

spk_0 environments, giving him unique lens on how to quickly evaluate technical visibility,

spk_0 taste ideas and iterate toward tribal product concepts.

spk_0 Daniel, we are thrilled to have you here for our season finale to explore this fascinating

spk_0 intersection of AI and robotics.

spk_0 Welcome.

spk_0 Thank you.

spk_0 Mack and Sam, it's an honor.

spk_0 Thanks, Daniel.

spk_0 So let's start with the big question.

spk_0 How do you see the relationship between AI and robotics evolving today?

spk_0 Are they developing in sync or is one advancing more rapidly than the other?

spk_0 You know, I think that AI has certainly been an accelerant for robotics, and especially

spk_0 in recent years, both for capabilities and also for the reduction of cost and the accessibility

spk_0 of these platforms.

spk_0 Project that will come up a few times today, I'm sure, is Huggings Faces Lay Robot, but

spk_0 you also have Nvidia's platform, which covers a variety of aspects, simulation models.

spk_0 And ultimately, it means that more of us can participate.

spk_0 It will be less expensive to do so, and the robots themselves will be more capable.

spk_0 Fascinating.

spk_0 Moving to the next question.

spk_0 You've been deeply involved in hackathons and cutting-edge development with the community.

spk_0 What's one recent breakthrough in AI-powered robotics that really excited you and why?

spk_0 You know, this is, I'll play on this for Huggings Faces Lay Robot project as well as Nvidia's

spk_0 N1 group model.

spk_0 One of the most exciting things, we hosted a hackathon, I guess it was just a month ago

spk_0 now, a little more than a month ago.

spk_0 And we were working on this open source project.

spk_0 Anybody could 3D print the parts in a weekend.

spk_0 You have to order some supplies from overseas, but assembling the robot only takes a few

spk_0 hours from you have no idea what you're doing to being able to put this robot together.

spk_0 So having a low-cost robot that anyone can participate in is really exciting because

spk_0 this used to cost many thousands of dollars, in some cases tens of thousands of dollars.

spk_0 And in the last couple of years, that's dropped to just a few hundred dollars.

spk_0 And I think the most exciting advancement was the release of Nvidia's open source model,

spk_0 which was very, very capable, very powerful, quite small, only a million and a half parameters.

spk_0 And three weeks after they released this groundbreaking model, they released a tutorial on how to

spk_0 post-traine with this very inexpensive robotics platform.

spk_0 So you have the most advanced companies really building at the bleeding edge, but just within

spk_0 weeks of release of the research, you also have instructions and the guides and everything

spk_0 necessary for you at home to do this.

spk_0 That's three weeks to go from emerging research to anyone being able to approach and access

spk_0 this.

spk_0 And I think I don't know how I would describe it exactly, but this idea that we have

spk_0 bleeding edge research, that previously was locked behind the labs of well-funded institutions

spk_0 and experts that had a depth of experience in the field.

spk_0 And now anyone in a few days for a few hundred dollars can access really the latest in what's happening.

spk_0 That's fascinating.

spk_0 And I think that's kind of like a re-emphasized the idea of open source model.

spk_0 Due to the open source model and also that robots being cheaper than it used to be, is literally

spk_0 helping these robotics fill the industry progressing further.

spk_0 So just going back to what you said earlier that in the AI is also powering the robotics

spk_0 industry, it's kind of like going hands and hands, and both of them are mutually benefited

spk_0 from each other from this journey.

spk_0 So thanks Daniel for your perspective on that.

spk_0 And for the next segment, a real world AI robotics application section, I'm going to pass it over to Mac.

spk_0 Thanks Sam.

spk_0 Daniel, thank you so much for the explanation.

spk_0 So what it came to know from you that right now it's so easy to, if someone has genuine interest in robotics,

spk_0 and from just an idea of face to executions, just in a matter of a couple of weeks, three to three to four weeks,

spk_0 I think someone can assemble a robot.

spk_0 That's actually a game-changing thing.

spk_0 Even three years back that we didn't even think of it.

spk_0 We're in that phase right now that one someone can literally build a robot.

spk_0 That's a great thing.

spk_0 That's the technology advancement that we have right now.

spk_0 So my next question to you.

spk_0 So I'm going to talk about now is more of the real world AI robotics applications.

spk_0 So let's get into the practical applications.

spk_0 So can you walk us through a real world robotics project that successfully integrates AI in a meaningful way?

spk_0 And I am particularly curious about your journey conducting the recent AI hack-out on important issues with Huggings this.

spk_0 I'll bring both of these together.

spk_0 I'll highlight a real world project that was one of our participants, one of our competitors in that competition.

spk_0 So the idea was you have a robotic arm that is capable of being controlled by an AI brain.

spk_0 And if you listen to people having conversations in the industry today,

spk_0 they talk about AI robotics being basically two pieces.

spk_0 You have the physical robot itself, and that's a whole domain where you're developing the physical bodies of these robots.

spk_0 And then you also have the brains of the robot, the AI that's thinking about the movement of that physical robot.

spk_0 So if you have that ability, and you can tie in other unrelated aspects of AI such as computer vision,

spk_0 you can utilize the ability to decipher what a camera can see, to process that information,

spk_0 and then to have the robot act physically on that.

spk_0 So one of our competitors had an idea of diverting recyclables from a waste stream.

spk_0 So you have a conveyor belt full of trash, but some of that trash is reusable.

spk_0 It's recyclable.

spk_0 So instead of sending it to a dump, if you could say pick out parts that could be recycled or reused,

spk_0 and divert them to other areas, you would improve kind of the whole of our experience.

spk_0 So the idea was you had a camera that was trained to identify recyclable material versus waste.

spk_0 And when it identified a recyclable item, it could then reach in, grab that recyclable item from that waste stream and divert it to be reused.

spk_0 And this was a high school student who had never trained an AI model before.

spk_0 They had never interacted with robotics before, and in that one weekend, the whole event was maybe 30 hours.

spk_0 They went from a pile of parts to assembling this arm, to going home, and overnight training a system to identify those recyclables.

spk_0 And then to have the robot then understand how it could pick up those items and divert it from waste and prevent it from just going into the landfill.

spk_0 Amazing, that's great.

spk_0 Great, great.

spk_0 And what was your experience with your hackathon partnership with PuggingFace?

spk_0 We can just elaborate a little bit.

spk_0 I think HikingFace is such a unique company, and they're so important for where we are today in machine learning and AI.

spk_0 They're the GitHub of AI models and data sets.

spk_0 And they don't just talk about doing interesting things.

spk_0 They actually invest in releasing projects that are ultimately useful.

spk_0 And this project has a 3D printed physical arm, which I have here.

spk_0 I'll show you the arm here.

spk_0 So the way you control this is you open and close the jaws of the robot like this.

spk_0 And then there's another robot that responds to the motion of this robot.

spk_0 Well, this was designed by a world-class robotics designer.

spk_0 It's open source.

spk_0 You can 3D print it in a couple of days on a readily available commodity printer.

spk_0 You don't need a special printer.

spk_0 It's just a few dollars worth of plastic.

spk_0 You can order all the parts from a list that they provide.

spk_0 And then you can assemble this on your own.

spk_0 Well, the only reason all of that came together.

spk_0 And the software was also provided to control this and record the training set and everything else.

spk_0 That was entirely because of HikingFace.

spk_0 So they made all of this available and accessible because they wanted to make robotics more approachable for everyone.

spk_0 And probably this wouldn't have happened in this way without HikingFace's involvement.

spk_0 So to celebrate a new design that they had, a new design iteration of this physical robotic arm, they hosted a worldwide hackathon.

spk_0 They had over 100 cities that participated, thousands of participants, and they let different individuals host locally.

spk_0 So Denver was one of those hundred cities.

spk_0 And we did it right here in Denver, Colorado, and had people come together learn how to build this physical robot, learn how to train it.

spk_0 And HikingFace was, it was just fun.

spk_0 It felt like we were working with friends and having a good time exploring new technologies and learning something that was exciting and interesting.

spk_0 I mean, none of us know what we're doing.

spk_0 It's too new for anyone to know what they're doing.

spk_0 So to have partners to share and explore and chase down whatever's coming next is just an exciting thing.

spk_0 So we're very thankful that they put this event on and allowed us to participate, supported us as well.

spk_0 Daniel, both of your examples are very fascinating.

spk_0 The first one was to differentiate the recycle from the waste frame.

spk_0 And the second one is creating a robot, you know, and then a survey short time.

spk_0 And they know like through HikingFace, for all these things.

spk_0 So can you walk me through like, let's say if I want to create a robot to do my landscaping stuff to my home.

spk_0 Then why it will be the initial step? Let's say I do not know anything at all, right?

spk_0 I'm totally no-wise and I don't know anything at all.

spk_0 So from start to finish, can you walk me through the steps and how even I can learn from HikingFace, the models and then just on a very high level?

spk_0 Yes. So I'm going to describe this in two ways.

spk_0 I think one is your curious individual who wants to understand how all this works and get your hands dirty and get involved, even if you don't have a depth of related experience.

spk_0 And then the other aspect would be say if you're a company that's developing a product or a service, maybe a solution that you would provide to the marketplace to do your landscaping, right?

spk_0 So there's a difference between the commercialization of these applications and understanding the raw kind of foundational pieces that bring all of this together.

spk_0 So I'll start with the individual aspect of this.

spk_0 Let's say you've never done anything with robotics before and you heard about HikingFace's open source robotics projects. What would it look like for you to walk through that process?

spk_0 Well, the first thing is you would need to download the model or find someone that has a 3D printer and is willing to 3D print the model for you.

spk_0 This takes a couple of days for the prints to run and then you have a pile of parts.

spk_0 You'll spend an afternoon assembling those plastic parts and you'll have to include the electronics, the motors that control the movement of the robot itself.

spk_0 And all of that costs you a few hundred dollars in a weekend if you do it yourself.

spk_0 Now you have a physical robot. Well, you have to understand how to move that robot, how to control that robot.

spk_0 And that's where HikingFace's La Robot comes in. It's a framework that has all of the software for understanding the movements and controlling the movements.

spk_0 But it also allows you to record the data for the task at hand.

spk_0 So let's say that you want to pull weeds out of your landscaping.

spk_0 In the same way that you have a camera that can identify recycling versus trash, you would have a camera that could identify a good plant versus a weed.

spk_0 And maybe a third party would provide that to you. There'd be a service, a plant identification service where you'd send an image from your camera up to some company and they would tell you, yes, that's a weed.

spk_0 You should pull that out or don't touch that. That's a good plant that you want to keep. Or you could learn how to train a model fine tune a model yourself.

spk_0 And you could actually use the open source AI that's available from HikingFace. And you could train your plants and your garden to identify what was weeds versus what was not.

spk_0 Now, that's a much deeper, more involved task, but it's possible for you to consider going down that road.

spk_0 And then you have the movement of the robot as well. Well, I showed you the robot earlier that is a handle that you can move around.

spk_0 What that allows you to do is it allows you to move the robot the way that you want to move it. And you record that movement.

spk_0 And then you tell the AI that based on what we can see, this is the movement that I want you to do. And it can learn from you.

spk_0 So you're actually physically moving the robot the way that you want it to move. And then it can understand that and learn from that.

spk_0 So that's kind of the introductory aspect of this. And you can you can take it as far as you want and go as deep as you want.

spk_0 You could train your own model based on your entire yard. You could put your robot on wheels so that it could travel around the yard for you and do these different things.

spk_0 And that's if you're doing it yourself. Now, taking that to a commercial application, there's many other steps involved in that.

spk_0 And if you maybe aren't as interested in doing it yourself and you want to leverage models that are solutions that are already out there.

spk_0 We have more generalized models that are coming out where you can even ask the model to do something for you.

spk_0 So you might have text that says remove the weed from the garden. And it will understand that text and know how to translate that text into the movement of the robot.

spk_0 Some of these things are becoming available at a research level today. And without question, we will see many companies that offer commercial solutions that will do these types of things for you.

spk_0 It's kind of to be just to be seen if they will make it available directly to us.

spk_0 Or if they will offer a product where it's an entire robot kit that will do weeding for you and you must go through their proprietary system in its entirety.

spk_0 But all of these things are going to be available. We're going to see everything from that raw open research all the way up to proprietary models and even proprietary complete systems that perform a specific task that you will purchase that for.

spk_0 Wow, that's a great way to explain things. I really like to learn so many things. I mean, just on your explanation.

spk_0 So this just came to my mind that I literally spend every weekend almost like a couple of hours sometimes or maybe twice in a month at least just to pick up the weeds from my grass.

spk_0 And I would be gone and back here. So David 11, that is something that probably you know like maybe I mean, I would have tried.

spk_0 Thank you so much for your explanation.

spk_0 You can always invite Daniel to your place and Daniel will feel the reward for you.

spk_0 Absolutely.

spk_0 There will be many people doing these tasks to teach the robots how to do them before we have the robots doing them. Yes.

spk_0 Okay. And and even going before going to the next segment, I have this one more question. So let's say how is enabling robots to adapt to unstructured or unpredictable environments like disaster zones, out of space or dynamic factory floors.

spk_0 What makes the difference between a traditional robot and an AI part one in these scenarios like that?

spk_0 Oh my gosh. So I think what's what's important for the audience to understand is in traditional robotics.

spk_0 There is an unbelievable amount of work that goes into the motions and the movements of the robot.

spk_0 And the amount of training, even if you have AI and some of these kind of record and playback types of approaches that are helping.

spk_0 The movements are very, very complicated to get right into to act in the way that you want them to act.

spk_0 And if anything changes, even the slightest amount of something changes, it can really be problematic.

spk_0 I'll give you a real example of this.

spk_0 With the new models that have vision and can see the, you know, there's a scene that it can see and interact with.

spk_0 If you record that with sunlight and the time of day changes, the changing of the angle of the sun or the introduction of new shadows can be enough to completely break the model.

spk_0 So you need a robustness to go into say a disaster setting where you have no idea what you're going to encounter.

spk_0 And you're not going to get that robustness from, we'll call it lesser models that are even available today.

spk_0 But if you go back to robotics just a few years ago, the idea that the robot could understand the scenery and recognize how to respond with an arm that might reach out and do something, let alone wheels that would have to climb over rubble that was strangely shaped or maybe slippery or whatever it might be.

spk_0 These are really, really hard challenges.

spk_0 But just in the last few months, we've seen some unbelievable advancements that are allowing these physical robots, which are AI enabled to respond to environments in ways that are that have never been seen before.

spk_0 It no longer is requiring these extremely structured, expected environments.

spk_0 We're now starting to see robots that can respond to unexpected situations.

spk_0 If they're picking up an object and moving it to another destination, that object can be oriented differently.

spk_0 It can be in a slightly different location and it can understand I need to move a little bit to pick this up and make it happen.

spk_0 So this physical intelligence that is being added into the robot is something that has a really dramatic effect on the types of things that were able to do.

spk_0 And just a few years ago, this was dreamy to think about the possibility of having a robotic system that could sense and respond to uncertainty.

spk_0 Great.

spk_0 Awesome.

spk_0 Really appreciate that the way you explained and for the next segment for technical foundations and challenges, I'm going to pass it over to Sam.

spk_0 Thanks, Max.

spk_0 Daniel, my next question is kind of piggyback on what you mentioned about the image classification part.

spk_0 So, I mean, if I go back to like five, six years back, I guess we would probably see the computer region book over there.

spk_0 So it was, it was a, we used to live in a world of you, like you only look once the image classification, like imagine it or you know all kind of models like daughter faithfully kind of like started in her, you know, Stanford lab, I guess.

spk_0 So now, you know, if you move to the track robotics transition part, like so what role this techniques like reinforcement learning, computer vision player, image classification playing the modern robotic system.

spk_0 Is it kind of like the same the way we've seen computer vision through the lens of a computer kind of like play the same way when a robot, you know, utilizing those same computer vision model and kind of like trying to do the image classification or categorized or even object detection in a dumpster or any trash, right?

spk_0 They're trying to find the, you know, the recycle items.

spk_0 So how could that set power?

spk_0 Yes, absolutely. I've got great answers for you here because the recycling the trash computer vision was actually yellow is what they used to do that.

spk_0 And also, a robot for for managing managing the video feed actually uses resonant.

spk_0 And so we always love to hear references to Dr. Dr. Faith Ali. So it's it's certainly certainly all of these things are are not are not like this sudden burst of an invention that someone just came up with overnight that is wildly different from what we've known it's truly building upon things that have taken decades to put into place and understand well.

spk_0 And that barrier of entry is just so much lower right the you know the idea of you know our our friends the high school student who who did that project the idea of someone who had never done computer vision before who had never trained a model before or fine tune the model who had never done anything with robotics to go from literally nothing to have a model that they had trained overnight and basically 24 hours.

spk_0 Like you can't even imagine someone being able to do that five years ago right so so the the barrier to being able to do these exceptionally powerful things is just is just so low it's it's such such a such a ease of access and ability to do some really incredible things and it's all built on the shoulders of giants.

spk_0 These luminaries and others who have who have spent decades building foundations for us to be able to do these things today.

spk_0 Yeah no it's fascinating in a scene this progress especially you as you mentioned that competition power it's not seems like the bottom they can more in a like somebody can even to

spk_0 train the model overnight with the limited resources they have this really great progress that we made in the computational world in especially in terms of computer vision you know and I think that's but you know like still we always take things in the grain of stalls.

spk_0 Especially when it comes to technology innovation there's always a room for improvement right so keeping that in mind what are the key challenges or limitation holding back wider adoption of intelligent robots in home healthcare or industry I'm not talking about Tesla robot with the very expensive is coming out for the next few months but in your opinion what you think that's going to still hold us back to implement that at large scale in household.

spk_0 Oh my gosh you know I think I've given hugging face in the learn about project a lot of our time and I'm going to keep doing that they they understood I think at least as far as I can tell earlier than anyone else that there was this opportunity to bring this technology to the masses.

spk_0 And that's why they put this project together you mentioned Tesla you know one of the one of the people that's involved in that project came from the Tesla humanoid project and they're now creating these robots and these processes that make it available for us to interact directly with.

spk_0 So we're seeing this incredible confluence of not only the these platforms becoming extremely inexpensive but also simplified in a way that almost anyone can access this I'm not exaggerating when I say this if you create a hugging face account run run a pip install then you know set up your environment run a pip install and run something like six Python commit.

spk_0 And you'll have created a data set that you can then use to fine tune a robotics model right and nobody bothered putting together something like that to make it easy for everyone to access now it's still a little bit of a technical challenge but when you when you reduce that barrier you now have people who maybe are traditional technologists or even people who are curious and are maybe a maybe ambitious enough to try to say I can fumble through this.

spk_0 And figure this out or even more traditional robotics people who are involved in the physical side but want to kind of dip their toes into the AI and machine learning all of this.

spk_0 So you have will call it the the non industry the non technical or you know non machine learning people getting involved in all of this.

spk_0 And it's very inexpensive and it's very approachable so that's one piece of it in addition to that you have in video really saying you know the latest in greatest in world class robotics movement models and then three weeks later providing everything necessary to apply that model to this same platform.

spk_0 So what used to be this massive massive gap between industry and research and everybody else is now this very very narrow gap you need a few hundred dollars and a weekend and an awful lot of curiosity and you're sitting at the table with everybody else that's doing all of this at the same time.

spk_0 You mentioned reinforcement learning there's new techniques that are coming out there's new research that's coming out there was a small visual language action model that hugging face put together and the entire thing this runs on commodity hardware so you know your laptop can run a visual language you know you can your camera can identify what's in the scene and you can say hey robot go pick that weed for me and it will do that.

spk_0 They trained that entirely on community data. What I mean by community data is those six Python commands that I talked about being able to record that data set well if you saved that publicly to hugging face.

spk_0 So that's the last year that was used to create this industry level model that could run you know visual language on on a laptop that could have been your data from just somebody that was curious about robotics on a random weekend.

spk_0 So narrowing this gap on the technical side of things where you have the actual implementation and the learning and everything else you're going to create people who understand these things in a way that allows them to get their first job in robotics no longer do you have to have an advanced degree in robotics to be valuable for a company that might want to implement some of these things and undoubtedly we're going to have companies that create platforms that make it easy for workers to access all of this.

spk_0 So you'll need people who understand how to interact with this on a basic level. So as the price comes down as the availability comes down as these more advanced opportunities are more widely available and accessible we're going to have these things just become part of our daily lives and I think I have to credit gents and Huang for this but everything everything that moves will be automated with robotics is is is where the target for this is.

spk_0 So the opportunities that exist the only reason it hasn't been done yet is because it's not easy enough for it's not cheap enough and things are becoming very easy and very cheap even for the most advanced of the technologies that are out there.

spk_0 Love that Daniel it seems like as the price going down for all this hard words and even though you have easy access to all the other model that you can even like run on your local computer.

spk_0 We don't even like a computer degree or probably like a high-end fence to body's degree I think at the end of the day just to back to that question I think the main challenge and limitation is that how much you're curious to learn about robotics and what to develop it.

spk_0 I think that's probably something that's going to hold us back in the future I guess technology won't be a limitation anymore and also to put a cheaper than I do is to be back in the days to build your local or something only an enterprise company can do but now everybody can do probably they're back here to put it in an inter-arge space.

spk_0 Yeah that's great thanks thanks and I'm pretty sure these winners are really interested to know more about robotics stuff and we have a few more questions to cover.

spk_0 Before I move to the next section the last question I do have for this one is that you mentioned about some of the technical hurdles that you've seen well you guys are working on the hackathon and stuff and I know they're like high school kids like

spk_0 we've really went back to go train the model overnight and came back next morning and polypore that report by itself. So in that particular you know going through the experience is there instead of like a technical hurdles that I school kid or any other attend this phase

spk_0 while you know going through the hackathon event I'm pretty sure some of our listeners would love to now go back to their homopies maybe try hands on or I mean so what are the tips you can share with them especially the technical issues people face during the hackathon and they can share.

spk_0 Absolutely yeah you know we had we had everyone from you know people who had never you know high school students who had never done anything with robotics before all the way up to people who had you know

spk_0 who were serial entrepreneurs that launched robotics companies so so there's a breadth of experience that's participated and these types of events and I think if you were trying to do this on your own there's a there's a broad range of domains that have to come together to make all of this makes sense.

spk_0 And none of it is none of it is very difficult or at least for the robot project none of it is very difficult but it covers a lot of ground you have the physical robots I mean you have to you know pick up a screwdriver and put it together you literally have to assemble it and 3d print it not everyone has that knowledge or you know there maybe they have a screwdriver buried in their junk drawer at home or something like that but they don't they don't do things physically.

spk_0 You may have the compute environment right just getting a Python environment put together can be daunting if you've never done that before I mean getting dependencies put together slow people down well if you're a developer that's not a big deal you've done that a hundred times or a thousand times but for someone who's never done anything in that type of environment there's a little bit of a hurdle to overcome.

spk_0 And with hugging face you have open source AI models you have the data sets you have the data collection you have the physical movement of the robot and I could probably come up with a few more domains here but the point is even though in isolation any one of these things is not really difficult it's a it's a it's such a broad range of domains and technologies that have to come together software machine learning open source platforms the physical

spk_0 robots and and if you don't have a strength and any one of those areas you kind of have to start from scratch and go figure that out and go do that so I think a barrier is certainly not having a breadth of understanding to help fill in blanks because none of us have complete strength across the board so being able to be more generalized more general

spk_0 generalists across the spectrum of domains that are necessary to implement these things that can be a challenge but at something like a hackathon where the stated intention is to learn together it helps it helps overcome that because you're offsetting your individual limitations by by the herd the groups complete capabilities.

spk_0 So I may be really good at the software but someone else has done the physical robots and someone else has a machine learning background well now all of us coming together is very helpful so.

spk_0 So I would say with that said we really didn't have technical hurdles that were highly problematic it was just for the inevitability that someone would have areas where they were less strong and we would just have to lift them up a little bit to get them where they needed to be but that's that's what we're there to do.

spk_0 That sort of hackathon is all about right we just teamed up you know we just trade up our knowledge and try to just share our knowledge and expertise to build something together.

spk_0 You know love that Daniel thank you for sharing your wisdom about all these robotics that are visual listeners and now literally interested to do some holiday probably the oldest that is listing down some of their project that I try to build robots that they are going to help you know use for their running their house.

spk_0 So also chose our kind of stuff now we have to move to this next section which is very important as well in that AI and robotics landscape which is it is 30 and social impact and to continue with that section I'm going to pass it over to Mac.

spk_0 Thanks, Sam. Daniel really loved the conversation the way it is going we learned actually a lot of things so the I mean the best part is that even I remember that you know like Tintual we're back that when we are in college we had to go through one to full semester to go through you know like the whole robotics course I mean to learn the you know all this stuff like robotics.

spk_0 And at the time it was coming into the market and you know so if someone wants to start something then they will have to go through forever you know like six months or one year preparation just to learn I mean robotics now the way we explained in just to you know create an account in hugging face and then just you know like download one AI models and then you know put six commands in Python and then you know like and then you can just simply train the model like in 24 hours or 48 hours.

spk_0 And then you know you can get that robot I mean this is really so much fascinating I mean this is a game changing so I don't think that technology would be a barrier anymore I think you know I'm someone doesn't have the money I think it still is actually able like you know with everything so thanks thank you so much for sharing that or my next segment is as robots become more autonomous how can we ensure their decisions remain interactable.

spk_0 It's a particularly aligned with human values. This seemed like critical challenge as we scale these systems.

spk_0 You know it's extremely important topic and we all should be a part of the conversation around this because there's a lot to it and our voices are important.

spk_0 I saw something in the last few weeks which I think is worth worth repeating here and it was a slide from a Microsoft presentation if I remember correctly from 1979 and it said computers can never be held accountable and for that reason we must never allow them to make decisions.

spk_0 So when we talk about a robot being autonomous I don't really believe that we would ever have a situation where that robot itself would be its own stand alone entity that could make its own autonomous decisions we should never allow that.

spk_0 What we need to keep in mind is that this is not an alien species that has found our planet and is now assimilating with our population.

spk_0 These are devices that we have created. These are our inventions and because they're our inventions we get to decide what they do we get to decide how they think we get to decide how they react to things and we must remember that behind every

spk_0 the wildly impressive autonomous robot is a is a decision that was made for how that robot would behave.

spk_0 So there's a human that's ultimately making those decisions and if that human says I'm going to delegate my decision making authority to this robot.

spk_0 We can't simply say that that human has now absolved themselves of any responsibility related to that robots actions because it was the human's decision to do that.

spk_0 So the people the companies those that are enabling and empowering these robots have a huge responsibility to do so in a way that does not impact our human experience in any negative way whatsoever.

spk_0 And if they do they need to be responsible in the same way that they'd be responsible if they themselves as a human were doing those tasks or doing those activities.

spk_0 The the buck has to stop at the decisions that are being made for how these behaviors happen.

spk_0 We can never delegate that to some seemingly autonomous device because we created those devices they are not standalone entities and they never will be.

spk_0 Great.

spk_0 My next question would be one of the very important and I think one of the most crucial aspect actually so let me give you the context a little bit so I was watching an European movie and then there was some of the construction workers like they were making when.

spk_0 Constructions and they were being replaced by robots.

spk_0 Even though you know like the lawn mowing systems nowadays I just saw that two years back in my office there were you know.

spk_0 Then people they just came and then they were just doing the lawn mowing stuff with the lawn mower and now I think last year I saw that there is only one guy came and you know put some robotic stuff and then you know I mean that was creating the inter lawn mowing for you know couple of acres land.

spk_0 So there is a growing debate around AI enabled robots and job displacement now or see your perspective on how we can balance innovation with social responsibility so in one hand we're seeing the advancement of technology but in other hand that we are seeing that you know job displacement.

spk_0 So how do you navigate these transition these transition part fully what do you think this is a this is a big conversation that has a lot of a lot of pieces to it.

spk_0 I think inevitably we will we will find that these devices are in some cases replacing human work or displacing human work.

spk_0 And certainly in the short term that's that's that's bad if you lose a job to a robot that's a bad thing because you need that income to be successful as an individual.

spk_0 And if you you know if we look at something like construction yes there are people who are highly trained and specialized in construction but there's many other people who do things like construction jobs because they're they're easy to get.

spk_0 And that may mean that we're taking jobs away from people who particularly need those jobs because they have other limitations related to what they can do.

spk_0 So so that's a bad thing right we can say it's it's bad for the people being displaced losing opportunity if they have to go find other jobs and they have limited skill sets they're they're going to be high they're going to be more competitive.

spk_0 And that's not a good thing so I think it's important that we are not just aware of these things but that we have some recognition that some type of contingency is necessary to accommodate the societal disruptions that you know could occur as a result of this.

spk_0 I don't know exactly what that answer is or what that looks like but there's lots of examples today where we have programs for people who have lost their jobs.

spk_0 We have programs for people who are in dire financial straits we want to keep people fed we want to give them medical care we want to do things to make sure that we don't just discard swath poll swaths of humanity because they're not in a good position.

spk_0 We hope that it's a temporary position and we find alternatives for them.

spk_0 So I think similarly to the way that we'll call it high risk environmental operations pay into systems that.

spk_0 That support any offset of any negative impacts that might happen if say an abandoned well leaks and needs to be cleaned up or something like that.

spk_0 We should be thinking about these types of disruptions and companies should be involved in that in some way what that exactly looks like and what it takes to get there I have no idea.

spk_0 But I think it's wise for us to recognize that there will be displacement.

spk_0 Now from a long term perspective jobs have changed all the time over the years the nature of work is wildly different than it used to be.

spk_0 You know I like to maintain a garden I like to get out there and grow some things but I don't think anyone listening.

spk_0 I would prefer to give up the grocery store for a life of toil in the fields to grow our own food.

spk_0 It's a huge amount of work but we don't have to do that anymore.

spk_0 And one of the reasons we don't have to do that anymore is because of the efficiencies that we've gained in the food system.

spk_0 Well these are basically more efficiencies that we're seeing and things like construction work can be very hard in the body.

spk_0 I personally don't know a lot of people who do construction throughout their entire life and are really happy that they did it for 60 years or something like that.

spk_0 It's just a really difficult job you're out in the sun you're lifting heavy things you're in a dangerous situation.

spk_0 Those types of jobs are great for an autonomous robot that effectively can't be hurt.

spk_0 You might lose a little bit of money if it breaks but you're not wearing out people's joints or sending them to the hospital or just figuring them because they got caught in some machinery or something like that.

spk_0 So there's a lot of really good use cases where we can use these devices to do the work for us that we'd rather not do.

spk_0 But the benefits of that in the long term we can't just discard everyone who's offset in the meantime.

spk_0 We have to recognize that there will be societal impacts and we have to respond to that appropriately.

spk_0 I love that. Adding one more thing, biggest retailer companies they have started on in their warehouse to packaging stuff to be done by robots.

spk_0 Even to send those goods commodities to the customer's house also eventually it will be by drone.

spk_0 So maybe this year's are coming in future that we might not need anybody, any humans in the warehouse maybe.

spk_0 They don't even drive any trucks or any transportation cars or whatever.

spk_0 It will be pulled through drone and then the drone will actually deliver those commodities to our houses.

spk_0 So these are I mean these are all speculations but these are also inevitable like in next few years.

spk_0 And having said my last questions in the segment from your work in both enterprise and prototyping environments.

spk_0 How do you see organizations approaching these ethical considerations in their AI robotic projects?

spk_0 I think as a general role the larger the company gets the less excited they are about small things.

spk_0 So you know it seems unlikely to me that say an enterprise would adopt a robot and have a working group for that.

spk_0 It's not a commercial robot.

spk_0 But you know it has physical limitations to the way it can lift the motors themselves while they are very durable.

spk_0 They're not made for an industrial environment.

spk_0 So there's a limitation to what you could do with it.

spk_0 And as a result I think you have large enterprises and larger companies that would say wait for some curious enterprise individual to come up with a new robot that they then commercialize.

spk_0 And five years from now the large enterprises will be happy to purchase those things and implement those things.

spk_0 As far as the kind of ethical considerations for that it really comes down to the organization itself and what what fiduciary duty that company has.

spk_0 If you look at you know traditional companies you know your your C-Corp that is publicly traded.

spk_0 Overwhelmingly their fiduciary duty is to their shareholders and their shareholders are going to vote at their annual meeting and the company has an obligation to fulfill that.

spk_0 And if the if the vote if the shareholders vote to say replace your entire workforce with robots.

spk_0 The employees of that company have a have a duty to fulfill that request.

spk_0 Even if it's against their moral or ethical individual beliefs.

spk_0 So shareholders have an opportunity to influence these outcomes.

spk_0 And you know the companies themselves are not always owned or influenced by shareholders.

spk_0 There's other companies that structure themselves in a way that give them more flexibility and what they can do.

spk_0 And you know maybe maybe we should be looking for companies that at least make declarations to say.

spk_0 You know we will not we will not fully displace humans you know or rationally or without do consideration right and and who knows what that means and what the application really is but.

spk_0 But oftentimes companies especially large companies are are beholden to their shareholders and that's kind of the end of their ability to act beyond that.

spk_0 Love that Daniel.

spk_0 And forward last segment in this episode the future vision and innovations I'm going to pass it over to Sam.

spk_0 Thanks back thanks Daniel I mean this this conversation has been shaping up really well.

spk_0 I would like to touch upon that you know that job disbursement thing you know I know we kind of talked about a few example like warehouse worker by the request by reports you know like

spk_0 majority of the physical activities might be an opportunity for implementing the bars that's going to probably help a human do not be involved in some sort of a high risk you know high life work.

spk_0 But what I was thinking you know yeah I mean obviously there might be some job disbursement will happen it would be a not like a full story if you don't think in that way if you think that in everything will be there human we still have their job even if reports can come and start taking over some physical activities.

spk_0 But you know in our opinion in some cases is possible that that the human who is going to probably doing a construction work will probably kind of like you know retool that you know like kind of like you know start learning about how to operate that report.

spk_0 There will be a possibility of that there will be reports coming in doing the construction work but you still need a human oversight to make sure those robots that bring their job are.

spk_0 Maybe you're going to think in that way that there will be more humans will be deployed to oversight being the supervisor of the reports not just doing that physical work by themselves maybe that's how we need to think about it and obviously they'll be a big back technology has been the most disruptive section director in last few decades is always in back in the society whether when the computer came the mathematician went on the parade and talking about their job loss and stuff we've seen that in the past.

spk_0 So I think we're going through the transition I just feel like we just have to think rationally and probably think little unorthodoxly like what we're going to do when machines are going to start taking over some of our physical or even Monday tasks I think that's probably the best way I guess Daniel you all be put that.

spk_0 So yeah so now since we're talking about the looking ahead and future how everything would look like it makes probably a few decades so I guess I guess the I mean the looking ahead you know like what trends or technology.

spk_0 So you know they should be watching in the next five years that will reshift you know AI and robotics landscape when we talked about computer vision is obviously was still using the similar kind of like computer vision models to train the you know AI part models and such but do you think there will be a huge shift the area where it's landscape due to the emergence of Jane AI element or there'll be something on the radar that's going to impact the human society or away of our work.

spk_0 Absolutely the I think something that I think about a lot is you know the the the non human solutions and what I mean by that are basically having having robots that look nothing like a human right and they're able to do work that a human would otherwise do but they're not doing it with two arms and two legs okay.

spk_0 I think maybe not a great example because there's a lot more other ones out there but you know you talked about drone delivery earlier well none of us humans have wings to go flap around and deliver you know packages single packages by air but we have this little device that can fly and carry a package.

spk_0 So that's one example of a non human solution and similarly jobs that we've created for humans to do our environments that were designed for humans to exist within their tasks and workflows that were designed for humans to do so the nature of work being human and having that change to non human tasks and non human work we're going to have specialized devices

spk_0 that do the things that we need to be done and they won't look or feel at all like humans so we'll have supply chains and logistical systems that it's that a human couldn't step into and take over and and and adjusts right because it'll be new systems that weren't designed by humans and cannot be managed or influenced by humans.

spk_0 None of us is going to strap on a jetpack to deliver a package if the drones stop flying right so so I think I think as a general trend we're going to see these entirely new spaces within industry that create these new types of works these new types of jobs that don't look anything like the human version that we've been used to for you know our entire existence up until today.

spk_0 So that's one piece of it the other piece of it is it won't be binary we won't just have no no automation and then we'll have robots there will be this integration that you talked about where in some cases it will be supervisors and other cases it will be augmented we'll have robots that we work with to extend our capabilities and extend our ability to do things I've seen these like mech suit.

spk_0 So I think that's one piece that people are wearing which increase human strength you know you could imagine something like that where maybe there's a third arm that gets attached to your back and helps you you know move some things behind you that you know you wouldn't be able to use so so there's many many different things that will be integrated into the human experience and also the non human experience will be an entirely new industry that will evolve as a result of all this.

spk_0 Just to submit it all in future the concept of iron band can become a reality absolutely absolutely.

spk_0 Thanks for thanks for your wisdom.

spk_0 We sure this is something our listeners will definitely resonate that and I guess now that's probably take me to the my final question of this segment is that there listeners out there who

spk_0 definitely want to jump on that board and start learning about how they can get into the field of AI and robotics so what would you recommend for their first tips to learn about AI and robotics and think about that there's some people out there who doesn't even know anything about AI and they also want to learn about AI and robotics both in tandem.

spk_0 Absolutely.

spk_0 But you might you might by now you might be able to guess what my response is going to be the hugging face lower robot project is an unbelievable starting point.

spk_0 It's very approachable it's very accessible it's inexpensive there's a community available there are people that can answer your questions and you can do literally everything from yourself which is very rewarding and also hugging face itself for machine learning.

spk_0 They have all the open source models and data sets and I think the robot project is really an exceptional starting point not only for robotics but also for machine learning and open source machine learning through hugging face in particular.

spk_0 It's it's the reason that I've been doing this for the last year is I haven't seen anything like this that brings together so many things in such an approachable way yet also has a depth available for as far as you want to take it.

spk_0 So I'm going to go look at the hugging face lower robot website join their discord start asking questions order a kit you know or figure out a build it yourself and just just be curious show up with an absolute boat load of curiosity and there's there's enough support available to figure it all out.

spk_0 Thanks Daniel love that the way you said the big curious in a best the main thing you know as long as you put your thoughts in mind behind something you will definitely learn about it and for our listeners will definitely share some links to the hugging face robot section in the podcast distribution so please go and check it out.

spk_0 So that's all from my end and I'm going to pass it over to Mac to record this session.

spk_0 Thanks Sam so we talked actually a lot of stuff today so the first we started with the evaluation of AI and robotics real world.

spk_0 AI robotics and applications technical foundations and challenges ethics safety and social impact future future vision and innovation so and definitely that we also talked about like you know like technologies not the barrier anymore.

spk_0 I think you know someone who wants to start working in robotics I mean is you know like just a matter of few steps so this has been an imputable session.

spk_0 Daniel and the perfect way to close out our first season so where can our listeners connect with you learn more about brainwave collective and potentially get involved in the hackathon and projects you were working on.

spk_0 Absolutely love this very easy go to brain wave collective dot AI we have easy to find link that goes to join our membership.

spk_0 We just opened up membership for community and also inventor and investor access as well.

spk_0 We'd love to have everyone be a part of that and we also have a meetup group here in Denver that we get together with for the public events that we have as well so even if you're not a community member we also host some public events to learn about these things and getting involved in these things also but yeah you can email me directly to annual at brainwave collective dot AI but just go to the website and that'll take you through every day.

spk_0 Thank you so much and that wraps our enter first season of inside assemble AI from marketing to robotics we have export AI's transformative impact with incredible leaders implementing this technology strategically to all our listeners thank you for the journey.

spk_0 If you found value in our conversations please subscribe share and join us for season two where we will continue cutting through the AI hype until then keep innovating but keep it strategy.

spk_0 Mack and Sam will be back soon thank you.

AI and Robotics: The Future of Intelligent Automation

Interactive Transcript

Topics Covered

Related Episodes

Out on a Limb | Nov 2nd | Hillhurst United Church

How Do You Respond to Jesus? John 5:16–47 :Daily Devotional, Daily Bible Study

Episode 240: Otto Aerospace CEO Paul Touw talks teardrop, laminar flow, and Phantom 3500

Burnout, Breakthroughs, and the Power of Getting Back Outdoors | Gun Talk Hunt

Share Episode