Augmented 112 === Intro: Welcome back to the Augmented podcast. Augmented is a podcast for industrial leaders, process engineers, and shop floor operators, hosted by futurist Trond Arne Undheim, and presented by Tulip. In this episode, the topic is automation and robots, and whether they're actually stealing factory jobs. Our guest today is Anna Waldman-Brown, a PhD student in political economy at MIT's Department of Urban Studies and Planning where she researches emerging technologies, Manufacturing systems, and the roots of rising inequality. Trond: Anna, welcome. How are you? Anna: Doing well, thanks. How are you? Pleasure to be here. Trond: I'm excited. Let's, let's talk about robots. Let's talk about workers. Let's talk about what's happening in, in the workforce these days. You've been working on this for, for a little while. I, I wanted to just quickly see if I could summarize what you've been up to. Anna, you, you soon have three degrees from MIT? Anna: Yep. Very soon, hopefully. Looking at May. Trond: So that takes you through physics, through tech policy to political economy and kind of urban planning. You were a Fulbright fellow, I'd love to hear a little about about that. And then there was something in your bio about a lab for chocolate science, and I'm almost tempted to sort of do the whole podcast on that. And then you, you are very passionate about empowerment topics and you worked internationally with the Fab Lab networks, which I love to talk about. And you worked in Kenya and other. Uh, always with the idea of democratizing Manufacturing and, and today's topic, robots and kind of the MIT's Work of the Future study where you, where you had a role and generally, if I'm correct, you are kind of positioning yourself as an expert in Manufacturing systems more, more generally. Anna: Yeah, absolutely. What I'm most excited about right now is American industrial policy, cause we actually have it again. Trond: Yeah. Isn't that something? Anna: Yeah. Yeah. It's incredible. Yeah. So I was working, lived in Ghana, did my Fulbright fellowship at, uh, cume and Kwame Nkrumah University of Science and Technology in Kumasi. Where I was looking at the informal sector. So these are folks by the side of the road who tend to have very informal like apprenticeship style, learn from their, you know, brothers and aunts and whomever, how to basically build and maintain an entire mass transit system. And so they'll get these, you know, FedEx vans. And convert those if you've ever been, they're coming in West Africa. They're called tro tro. They're called matatus in East Africa, and they're popular across the Philippines, parts of India, south America as well. They're basically these vans that get retrofitted to fit about 15 people. And so they're these little, mostly privately run mass transit systems and they establish bus routes based on who wants to go where. Uh, you can sometimes ask the bus driver to go a little out of their way to drop you off if you've, you know, got a goat on your shoulders and need that goat to go home and maybe it's a bit more of a hassle to go around the other way. So there's this informal, but it's also very effective mass transit system that's a lot easier than what we have across most of the US and it's entirely maintained by these informally trained auto mechanics, um, who are extremely skilled. Don't always have the sa- most safest conditions, but they're doing things like using Soviet era lathes to build ball bearings. The other th- talk about chocolate, the cacao supply chain is also entirely built in Ghana around these old trucks, which are often also being maintained by, um, these informal auto mechanic networks. This stuff is fascinating and I see, you know, how you were talking about democratizing technology, taking technology into your own hands and retrofitting, and I guess that's the story of human in- invention, right? It's not just these technically educated engineers, you know, uh, who are working for the Pharaohs, you know, like in a stylized system, technology goes through all these life cycles. I find that very fascinating. Trond: Yeah, absolutely. And that that's how so many supply chains actually function. That assuming some stat that a majority of garments in the fashion industry go through a cottage industry at some point. Whether that's to sew on a button or like doing little embroidery steps, there's a lot of these, what I think of as the last mile tasks that can't quite be automated or that don't really make sense to centralize because this FedEx van is considered totaled in the US. You ship it as scrap metal to Ghana. They're like, yes, this will be the bus! This will travel along the main bus routes! There was an academic who called it a "state of maintenance through constant repair", how these buses continue running all the time. So I have to say it sounds fascinating obviously, but sometimes it does go, go wrong. I'm not gonna name the name of the company, but I was on a family trip to Iceland once and and used one of the companies that that uses much, much older cars. And the first car, you know, arriving there very late at night, I think it was like eight or nine, uh, raining, horrible weather. Get into the first car, just about to pull out of the parking lot and the steering wheel falls off. Anna: Oh woah. Trond: So I go back in and I say, Hey, the steering wheel fell off! So they're like, oh yeah, no problem, no problem. We'll give you another car. And then we get into the second car and it's the last car they have. And by that time, we have now taken a look at their factory and I see exactly what you're talking about there. And it is ingenuity and, and I, I think 95 times out of a hundred it goes really well. But anyway, so we get this absolute last car in, by this time it's like 10:00 PM. So I get on the road and it, uh, you know, the gas meter, like something, the gas tank is leaking. For the whole trip, we have to keep filling gas in these very sparse gas stations in the, you know, in the boondocks of Iceland. So anyway, I think it sounds fantastic and you know it, and it is such a great journey to tweak technology. But sometimes we don't realize, I guess if we're not living it, that it also has some drawbacks. For example, you know, the cars, then, that don't work as well and they have some issues. Anna, I wanna get to robots, but just let me understand. So the, the lab for chocolate science, was that literally talking about the science of chocolate or what did that have to do with the cacao supply networks? Anna: It's a student group that was started by, she's now a biologist and founder of a chocolate company, I think about 20 years ago now. So before my time that started out, this talented chocolatier was making truffles for all of her friends. And at the same time, MIT was funding student life improvement projects. And so she realized Ariel Segall, Ariel realized that she could get funding from MIT to teach her friends how to make truffles. And this became MIT was just throwing money at anything that would improve the miserable mental health of students as, as we still do. It's a wonderful place that also has some challenges, but she got a bunch of funding for it. Uh, at the time. This was before CSAIL, Computer Science and Artificial Intelligence Lab used to be called the Lab for Computer Science. So it was a funny joke at the time to establish the Lab for Chocolate Science and then would also host people giving chocolate science lectures. Um, I ended up working with a cacao growing cooperative doing technology policy in the Ecuadorian Amazon for a summer in undergrad. And, um, hang out with applied physicists who does literal chocolate science, who came and gave a number of talks on cocoa butter crystallization. There's a lot of genuine science that goes into it. We we're not publishing the research where it's mostly a bunch of undergrads and a couple grad students who come over and help out on occasion. Trond: I, I love that. No, I mean, it's interesting. Well, this doesn't get us any closer to robots, but I think there's this, uh, issue of the supply chain, obviously, of chocolate is threatened. And I know that you're part of, uh, some research here that's talking about how machine learning and other approaches can help climate change. And I think the chocolate supply chain is a vulnerable one. So there's uh, actually reasons beyond just fun to care about this topic. Both if you love chocolate and if you love the workers that are making a living out of it. So, interesting topic for, for another day, I guess. Anna: Yeah. Also very driven by informal sector, um, farming and processing as well. Trond: So let's jump then straight to the robots. So you wrote a piece in Wired a little while back, which, uh, I, I think was derived in work at MIT Work of the Future study where you, you said the headline was 'Why robots are not stealing factory jobs'. Can you just first take us to, I believe it was some Oxford economists who famously wrote that the robots were stealing factory jobs. Anna: Yes. Trond: Some years ago. I wanna just understand how did we get there to this being a problem? Because weren't robots a solution to certain problems? Now, how was it suddenly construed as a problem and what was your work in this regard? Anna: So early history of welding robots specifically were some of the first robots to be adopted with Unimate back in GM at a time of intense union conflict as well. And so there's always been this tension between, let's you know, get rid of the dull, dirty, and dangerous Manufacturing jobs by automating those, as well as this is due to the contentious labor relations in America, we saw this with, uh, Elon Musk and Tesla's Fremont factory recently. Americans have always been, in my opinion, a bit too overexcited about using automation to replace the skills of workers. And so you saw this with early welding robots where as soon as you automate the system becomes very rigid, right. You've written a lot about, in terms of the idea of Augmented Lean is flexible automation. And early automation was really, let's put in as many robots as possible so we can get rid of these welders or these skilled machinists who need to have enough skill that they still have a lot of power in the factory. And so gm, a lot of the American automotive companies. In the eighties, were really trying to use automation as a de-skilling tactic. And at the same time, of course, Toyota over in Japan was taking a, the opposite approach, was pioneering lean and really sort of looking at the numbers, you can see that Toyota was way more successful with strategic automation, and strategic augmentation of humans who have those particular skills, and really helping humans develop a more holistic idea of what's happening on the shop floor with quality circles, with changing around who does what tasks, making sure that the workers are more flexible, and whereas Americans, were using automation at the time, sort of by and large, there are a lot of exceptions, to sort of make these much more rigid replacements of tasks. And then the workers, this is what you see with the skills bias, technological change. Instead of having welders, now all you need to do when you've got enough welding robots is have someone who would do the materials handling. They'd put the part in, they'd hit go, they'd step out of the cage. The robot would do its thing, and you could have totally unskilled workers on the frontline. Trond: So Anna, lemme just play devil's advocate, which is tricky for me because I'm clearly on your general side of this argument. But just for the sake of it, what was it that these large American companies were seeing in de-skilling? I mean, are they seeing what we're seeing today, which is obviously skilling is still a problem in the Manufacturing industry and everybody is trying to figure out how to do this. Because how do you scale skilling? So if you have skills that need a change because your work content is still fairly advanced, or at least it's advanced enough that not everybody can just come off the street and, and then be a competent Manufacturing worker. You have to find a way to train and teach skills at scale. So, you know, devil's advocate, I mean, isn't this a good thing? Could they actually have succeeded somehow if they had done it somewhat differently? Anna: Yeah, and I think what you see now is the robots have become more flexible as well. This is a joke in welding that we've had a shortage of skilled welders for the past 60 years, right. You talk to welding workforce advocates and they'll say for their entire lifetime, there have never been enough welders. The last time we had enough welders in, thinking of the American context specifically, was during World War II. There was a huge initiative to both re-skill, this was a lot of, um, black workers and a lot of women, and it turned out women in particular who'd never worked in factories back in the forties were used to sewing a lot, and sort of had that fabrication expertise as well as the fine attention to detail. And so firms found that, this is Rosie the Riveter moment, that a lot of these women and other people who hadn't worked in factories already had a lot of the types of problem solving and attention to detail that would make them really good welders, metal fabricators, riveters, et cetera. But that was a monumental workforce campaign. As well as providing childcare, which is another main reason that we don't tend to have a lot of people in the workforce. The folks who dropped out during COVID often did so for childcare or caring for elderly or ill people, as well as some of the first firm sponsored healthcare policies were, were around that time. And at the same time, redeveloping the technology. And so the ships that were being built can go visit the Rosie the Riveter Museum in Oakland, California. And they have a great exhibit on how they redesigned these battleships to be more modular. And so there was a whole systemic approach to, let's figure out, not just how can we train people to tap into it, but how can we design these systems so that they're more standardized so the products themselves can bring in people who can get much more quickly up to skill. Trond: So you were counter posing American Manufacturing with Japanese, but there was also Germany, right? Very successful Manufacturing Nation. What is it that they're doing different in Europe, for example, when it comes to robots or, or automation overall in the work? Anna: Yeah, so I need to get back to you cause I need to go visit a bunch of more German factories with robots. Um, but next month that will happen, um, working with Fraunhofer IPA as well, which has the Future Work Lab. And I think because American factories were really burned by over automation, Americans have sort of started rethinking our approach to how we're bringing robots into factories. One of the real opportunities for automation now are these smaller manufacturers, small and medium firms, the SMEs, which are doing high mix, low volume production, and so the numbers are actually similar in the US and Germany. SMEs tend to have about 10% of them have robots. Trond: Let's stop there for a second though, because that's crazy, right? Because the whole MIT study on future of work started because everyone was saying robots are taking over. Anna: Right. Trond: And then what you found, I guess, is there are very few robots. Like, where are these robots that everyone is talking about? Anna: That is the, the subtitle of, of one of our papers. We started with a randomized selection of firms and the team in China actually found similar results. A bit more robot adoption in China, but looking at small and medium firms, if you assume the Oxford economists, Frey and Osborne are correct, I take a lot of issue with some of their methodology, but they say that something like welding is 90% automatable, and the way they do that is they break that down by tasks and look at for each task, what's the degree to which machine intelligence, plus robotics or whatever automation, physical automation you're using, can replace that task. Even if welding is majority automatable, that minority of tasks that are not automatable are going to be found disproportionately in these small and medium firms. Trond: Interesting. Anna: And so both in the US and Germany, um, and Japan, China, wherever. You go out, you're looking for small firms with robots, you're gonna have a really hard time finding them because the use case is just not there. Trond: Mm-hmm. Anna: Uh, even with collaborative robots, even with more affordable robots, with these no code user interfaces, one problem is that firms are still very leery of, oh, a robot, it's gonna come in, it's gonna be dangerous. We're gonna have to reorganize everything. We're gonna have to have standardized parts. And so there, there is a fear factor as well as a real practical fact that in the firm's, so we failed to find enough welding robots in our randomized study of Manufacturing firms in Ohio. So I've spent the past couple trawling the entire internet, looking for firms that have welding robots, which has been a real challenge. There are not a lot of them. And when you find these firms, you ask them, what percentage of welding jobs can you automate? And the answer, some of them have automated, you know, 40, 50% and most of them have automated, you know, 3 to 5%, maybe 10% of welding tasks. It really depends by firm. But the challenge with welding. Which I see as sort of a broader challenge as well is that you need to design a jig to hold this part because basically you don't have a smart enough robot grippers the way that you have a human hand. Right, hand can grab a piece. You can move this around as you're welding it with a robot, you're gonna need to stick that thing on a jig, and that jig is going to need to be designed so that you can reach all around. If you're welding, say the seams on my phone. Right. That jig needs to be carefully constructed to hold it so all of those seams are available and it can be moved maybe once or twice during the welding process. And so there's a whole added system that, I think this is true for automation as a whole. You need to make that environment more automatable and more friendly to robotic. Trond: Well, it strikes me that, I mean, the use case for automation generally is scale, yet we seem to be so eager to use it for something different. Anna: Yes. Trond: To use it when there is no scale involved. That almost sort of defies logic really, doesn't it? Because if you don't have hundreds of operations going on every day in that are monotonous Anna: Mmm-hmm. Trond: Then why would you even look for a robot to do that task? So I, I, I'm struggling even with, from a management perspective, I guess I understand why you didn't see robots in these smaller firms. Anna: Sure. But the cool thing is now you're starting to see them. Trond: Right, so, so tell me how that change is happening. Anna: So, yeah, obviously it depends a lot on the product. One of the challenges is firms that are doing, you know, even a batch of 50, if that batch repeats every so often, and so it becomes a supply chain issue. Can you convince your customer to place repeat orders or to buy larger batches? Then that would justify the investment of building a welding jig. The other challenge, there are some firms that are making hundreds of the same thing. Um, we saw this with an exhaust pipe manufacturer where they're just minor tweaks, and so you can program the robot to do something. You can make these minor tweaks. You still often need to build the whole jig from scratch. But their main problem was upstream. The exhaust pipes weren't being fabricated to precise enough tolerances, and the robot would keep getting lost on the seam. And so one of the things that we're starting to see, see now is better seam tracking software. And this is often using machine vision. Uh, or you have like little laser systems. There's a really cool startup called Path Robotics out in Columbus, Ohio that's building fully automated welding robot cells. Again, you still need the jig, but they're integrating it with these smarter jigs. You know, are basically a robot, and so the jig gets moved around on a robotic turntable at the same time as it's welding. The main advantage of this is something like utility poles, which if you're building this many meter long pole, it'll twist a little bit from one end to the other. And so you don't know where that seam is unless you have a robot. So basically you upload the CAD file first, and then this robot will project a little point cloud on the part, and so the machine vision system can figure out where exactly that seam is as compared to the CAD file. And then as the robot is going along, it'll periodically scan to just make sure, so there's sort of a triple redundancy going on here. So it'll periodically scan to see where that seam is. And so this is brand new technology. I think we're gonna start seeing more of this in the automation world, and that can be a different utility pole. Trond: And I'm, I'm curious, so you've, you've done a lot of research. You have been in small and medium firms in many countries, looking at the welding use case. What is your observation and perhaps advice? I mean, clearly if you're a worker, if you're a manager or if you're an executive, it sort of depends and your perspective might be different. And to some, obviously this is, well, it's expensive, right? If you're in a leadership, this is a big decision to automate or not. Uh, in specific use cases, if you're a worker, there are lots of other sort of issues related to skills, but also this fear of obviously being automated out, out of a job. With your best knowledge now, like if you are in the industry right now or thinking about entering the industry, you know, if you think about the worker or perhaps the engineer in this situation and the executive, those three roles. Anna: Mm-hmm. Trond: What should they be preparing for? You've watched this tension. You've watched this misaligned, non flexible automation that's now turning even in the US into somewhat more flexible automation, or at least the technologies are more adaptable. What should a worker be thinking here? Or are, uh, Frey and Osborne, the Oxford economists finally right, and they have to figure out something else to do? Or is it more that they just need to change the pitch or change a little bit their skills and they'll still be welders what's happening to their welding function? Anna: So thinking of welding in particular, this is why I'm excited about your concept of Augmented lean. There's a lifelong learning element to it. Where sort of at all levels of the factory, you want to really empower and enable problem solvers at every level of the Manufacturing organization. But what this means is you still need deep technical skill because in theory you'll need fewer welders, um, than we have in the past, cause we'll be able to automate more. But in practice, I've been chatting with a bunch of policy makers at the Department of Energy in the US now. If we're gonna have any hope for a functional green transition to get off of fossil fuels, we're going to need massive infrastructure, massive Manufacturing investment, not just in America, but all over the world. And in order to do that, we're gonna need a lot of the OEMs to figure out their large scale production. And all of these small and medium firms that are further down in the supply chain will continue to have a lot of variable production and a lot of these one-offs, high mix, low volume sort of things, where you'll still need at least a couple people who have in-depth technical skill. And so what I think is exciting about the collaborative robot space, particularly in welding, is you talk with these factory managers. They say that they've found, and robot integrators are now starting to recommend this. The best strategy is you get some young digital native who's maybe not that much of a welding expert, but really adept at smartphones. And can pretty quickly figure out what a tablet interface might look like, which is the interfaces of these collaborative robots or even industrial robots that, as you know with Tulip, you're developing all these no-code and low-code interfaces, that allow anyone without that technical knowledge on the programming side to come in and start creating killer apps or killer robot moves, whatever, you know, killer in the metaphorical sense, because these are all much, much safer robots, and so you don't need to stick them behind a cage. Now you can start interacting with them in person. With welding, obviously you still want the welding to be behind safe welding curtain and have your various safeguards in place, but you don't need that large footprint on the factory floor that your industrial robots used to have. And so once you're enabling this young guy, what everyone says is you want a young guy and an old guy working together and the old guy is gonna come in with the welding expertise. And so for one example, there's a small manufacturer we talked about in Maine. Where the older experienced welders couldn't find anyone, so they got this young woman who's, I think she was 19 years old when she finished a metallurgy bootcamp. It was nine months she'd never welded before in her life. She shows up at this factory and she figures out the robot within a couple days. Because it's one of these collaborative welding robots that's super easy, low barrier to entry. She knows nothing about welding though. I mean she, she's done the welding boot camp, and so she's familiar with all the things, but she doesn't have that familiarity of how do you hold that welding torch? How do you figure out the angle? How do you have the feel, the tacit knowledge to it? And so even though she knows how to program the feeds and speeds on the robot, it's really helpful to have some older mentor who's also available to help her and guide her. And so what I think is most exciting now for both workers and for managers is that these technologies are enabling upskilling in a way that you come in without so much of that technical knowledge, but you're willing to problem solve, you're willing to learn. And so these young workers, this woman for instance, she figured out not only how to use the welding robot, she became lead welding robot programmer. She's now learning to weld even better because the robot is helping her improve her welding skills. And now she's going on to program the laser cutter because she's familiar with the metal fabrication now. Um, she's familiar with the machine interfaces and this isn't a quick boot camp. And so I think that is the way that, and again, you know, they're 90 plus percent of most of the welding jobs in small and medium enterprises are not going to be automatable for a considerable amount of time, and so you're still gonna need that welding expertise, but we can now start to mitigate some of these skill shortages with smarter technologies that can help train the workers at the same time as you have that mentorship of the experienced welders. So then Trond: if you go to the kind of the engineer, or perhaps it's the, it could even be a, a planner of some more, uh, managerial type. At the next level, the, the, the supervisor, whoever, whoever is guiding the tasks or structuring the tasks, whether they have an engineering degree or not. What is their situation then is in this landscape? Should they just try to jump on these no code interfaces or, I mean, what? What will their role be? You seem to have this contingent answer to it, like sometimes the 19 year old can rule because they come in and they have all the skills needed. You don't need the 10 years of managerial, sort of know-how, and other times it's not. So I'm, I'm just curious, that layer that's sort of in between layer, you know, with some skill coming into the job and some experience managing bigger tasks, more than one task, how does that function change, for example, and let's keep stay, stay with welding. What's happening in that layer? Anna: The ideal case is that the lower level robots become managers of the machines, but you're still gonna need someone to both know the welding skill and know how to do, there's still a lot of artisanal work involved in welding in different materials, different types of jobs. You might need to know the right angle, or you might need to have more familiarity with how the robots work themselves in terms of, you know, designing the right tool path because you can automate a lot of that and there is a lot more skill and systems being built into the robots, but you still can't quite make up for the expertise of someone who's been, say, working with welding robots for a decade. Trond: There still is a role for, for that layer in, in, in between. It's not like this is gonna very rapidly lead to the abolishment of, of managers or supervisors. There's, there's gonna have to be supervision on the shop floor still. Anna: Yeah, you'll need supervision. You'll need the experienced problem solvers who've got those decades of experience who can come in and easily look at something or listen to the robot. A lot of people will talk about, oh yeah, we listen for the welding robot to do something wrong. Obviously if you're a novice welder, you're not gonna know what a wrong weld sounds like to the same extent. Trond: That's fascinating. The listening part is actually super interesting. You know, there's lots of research on that Xerox part, for example, where they were doing these copiers, the copying machine repair people were, have also been listening to the photocopier for decades. What about the executive level in Anna? What, what happens there? Is it a case of just becoming a little bit more savvy and realizing that, you know, automation is not one size fits all? Anna: Mm-hmm. Trond: Or you know, how, how do you think about automation then when you're leading a company or, or advising? Sort of, you're in the senior management of a smaller Manufacturing firm. How do you think about automation? Anna: There's a firm we visited out in Massachusetts, one of the most thoughtful firm owners, I think. Who said, I've been keeping an eye on welding robots for a while and I see so many welding robots on the used machine markets on like the eBay for used industrial automation equipment that I am never going to buy a welding robot unless I can run those numbers and know that this is going to work for, in his case, extremely small batch Manufacturing. They do a lot of prototypes for startups. And so I think there is a real role for the executive to think about what types of products are you making and where can the technology fit in. And so we did see, we called them the doorstop robots. I think we saw about four or five doorstop robots in Massachusetts. Across like 15 robots, welding robots overall. So there's a lot of them out there. There's a lot of over enthusiasm about the robots. Often people will get these on sale. Toyota is selling off, so the larger automotive firms will entirely redevelop their automotive lines. Six to 10 ish years, six to eight ish years. And so they'll sell those on, they use machine markets and small firms are like, yeah, robot. And they get one. And I've been really impressed by, particularly with the newer robots, the number of firms who have no automation experience, but deep fabrication knowledge and they get a robot and oftentimes they can figure it out. But I would say about as often, either the job changes, and so a lot of the time it's the supply chain. Or they get a robot, they don't quite realize how inflexible they are compared to welders, even the newer collaborative robots, right? If your parts aren't fabricated up to spec, and this is something that people were really not realizing the importance of, and again, this is really valuing the skill of the shop floor worker. This one firm owner said, yeah, the, the sort of the buck stops at the welder. Because the welder realizes, oh, you know, there's too big of a gap here. I can just add a little more wire and, uh, no one will notice. Or there's, the welder, often has a hammer and will be hammering that part to fit a little bit better. And so there's a lot of this finagling at that end step, even with parts that are, you know, medium sized batches, right? Making dozens of these things. In theory, they could totally be made on a robot. But you need the materials coming in to be rolled to a uniform thickness, depending on your quality, right? You need the metal press operators to be doing that exactly where that crease is supposed to be. You need the fabricators to be folding it properly and every millimeter is gonna really add up. And so if your parts aren't fabricated to tolerances, you're not gonna be able to automate. I think firm owners sometimes will underestimate how skilled their welders really are at adapting to these sort of variable parts coming their way. And so there is really this role of conversation with the people on the shop floor, like, look, how much hammering do you need to do on these parts? And that's kind of the gauge to whether or not something can be automated. Trond: How much hammering, huh? I'm gonna move us just very briefly to kind of the future outlook. And I know you know you, you're not a futurist, you're a researcher. But I guess the question is, and let me just paint two pictures. One, one would be the picture that automation and robots are now coming because they're getting better and they will come. So then I guess my question is just when will they go fully mainstream and what will that do? And, and then the other scenario is what you were sort of painting with, with climate change and, and transitions of, of our industrial system. I'm imagining that they will look a little bit like your African transportation system. You know, there'll be an enormous amount of retrofits. You have oil rigs, you have electric infrastructure, and all of these are machine parts too. And they will have to be, like you said, hammered into a new mold, into a new system, and there's gonna be welders. And I just have a hard time, perhaps, in that scenario thinking how much of that could be automated because you are not then dealing with a greenfield as one calls it. Right? You're dealing with brownfield infrastructure. Anna: Yes. Trond: Where you need to go in with your eyes, presumably. Uh, unless, you know, robotic vision and, and computer vision has got much better than I think, and, uh, you know, you're somehow able to, uh, roboticize that entire transition. Where are we heading? Is it one or the other, or both? Anna: Someone was saying industrial transformations, they're not an event, they're a process. And so at the same time as you've got bullet trains or Hyper Loops, you'll still see these shared vans that are hacked together from old FedEx vehicles that are providing pretty effective mass transportation. So it's funny actually, how many times we've seen robot adoption or a huge push for automation when like, the daughter takes over her grandfather's firm, or the new owner comes in and he's, you know, been automating this other factory before he buys this legacy firm and comes in and brings all of this technology. And so there is a mindset shift that not necessarily the younger, but the more tech enthusiastic generation can bring. But there is also going to be an awful lot of hammering going on as we're retrofitting these newer factories. Uh, I think it's telling that Amazon, which has acquired number of robot companies for its own warehouses and other automation systems, they, at . Least last time I checked, their policy with brownfields was they're not worth automating. And so keep the older warehouses operating, try and retrofit those as much as possible to ideally make them friendlier to the human workers. But then it's really the greenfields when you're building this thing from scratch where you can redesign the entire warehouse to have your, you know, robots going around on sky cranes and what have you. So that vision for the future is much more applicable when you don't already have a lot of infrastructure and, um, frankly, we've got a lot of infrastructure everywhere. And I think it would be a real waste not to take advantage of what we already have built. I think some of the most exciting clean tech companies right now are even building up on existing infrastructure. There's like the iron air battery team that's coming out of MIT where. They're getting a product that usually goes into steel Manufacturing. They're getting these iron balls, and then they're using that as the feed stock for this battery that rusts and unrusts their iron, and that's how they store the energy there. But that entire company is premised on the idea that here are, these effective, fully functioning supply chains that we can just tap into and figure out how to build upon that. And so I think the equivalent in Manufacturing, where you figure out those supply chains, you know, we've got a bunch of rail infrastructure in this country, we can build better trains on top of these existing rails. But that does take a lot more actual human expertise because you can't be designing these systems from. That's not to say that the fusion companies aren't gonna be doing awesome work, and they're gonna be inventing entirely new supply chains and entirely new skills. But one of the things that's holding back Commonwealth Fusion systems, which is just down the block for me here, is they cannot find enough welders right now, and metal fabricators and skilled trades people, because that's just how stuff gets built. And so I think we do need a huge push towards workforce, at the same time as we're making these Manufacturing automation, these new technologies as user friendly as possible, so we can use the technology to be upskilling and augmenting those workers at the same time as you do need some amount of mentorship through the actual humans to learn those fabrication skills. Trond: Anna this is fascinating. So what I'm learning here is, you know, hammering and keeping the algorithms at the same time is, uh, is a good idea. Just keep your hammer, keep hammering, but you also need the, the algorithms, and the computerized reality is happening, but you we're gonna need both for the foreseeable future at least. And it's been fascinating. I feel this little dive into welding has given me some food for thought. Uh, it was fascinating to hear from you. I wish you best of luck in your future work with that and thanks a lot for spending time with us here! Anna: Thank you, my pleasure! Trond: Thanks for listening. If you liked the show, subscribe at AugmentedPodcast.co or in your preferred podcast player and rate us with five stars. The Augmented podcast is created in association with Tulip, the Frontline Operation Platform, connecting people, machines, devices, and systems in a physical location. Tulip is democratizing technology, but also importantly, empowering those closest to operations to solve problems. Tulip is also hiring and you can find Tulip at Tulip.co. Please share this show with colleagues who care about where Industrial Tech is heading. You can find us in social media. We are Augmented Pod on LinkedIn and Twitter and Augmented podcast on Facebook and YouTube. Augmented: Industrial Conversations that Matter. See you next time.