A Glimpse of Automation’s Future: From MRP to IIoT and Beyond
Inductive Conversations43 minute episode Apple Podcasts | Spotify | Google Play | PodBean | TuneIn
On this episode, we have the pleasure of bringing David Greenfield, Director of Content/Editor-in-Chief of Automation World/PMMI Media Group, together with Don Pearson to discuss the future of automation in manufacturing. David shares his insight on the evolution of MRP into what we know as ERP today. Don and David also discuss the rise of IIoT and the progression of cloud services, how Ethernet got a foothold on the manufacturing floor, and the great potential that mobile devices and mobile-responsive screens have to offer. David also shares his insights about a subject that is near and dear to him: robotics.
“Manufacturing is no longer a black box, so to speak, with IIoT ... It changes the thinking about approaching the specific granular aspects of manufacturing data.” – David
David Greenfield joined Automation World in June 2011. Bringing a wealth of industry knowledge and media experience to his position, David’s contributions can be found in AW’s print and online editions and custom projects. Earlier in his career, David was Editorial Director of Design News at UBM Electronics, and prior to joining UBM, he was Editorial Director of Control Engineering at Reed Business Information, where he also worked on Manufacturing Business Technology as Publisher.
Don: Well, good morning, David Greenfield. So glad to have you here this morning. I think I would like to start by just asking you to introduce yourself and tell us a bit about... just a bit about you. I know I know something, but the audience doesn't.
David: Absolutely. And well, first off, thanks, Don, thanks for having me on the podcast. Really great to be here. It's fun to be on this side of it, as opposed to the other side of the podcast I'm usually on. So yeah, I appreciate this opportunity. So I get to see it from this side. I guess just a quick rundown about myself: I am currently, and have been for the past 10 years now, the Director of Content and Editor-in-Chief of Automation World, which is part of PMMI Media Group, but I've been covering this industry, industrial technology, since 1991 when I first got into it. I got into this somewhat interestingly, I guess, that I didn't come from a manufacturing or an engineering background; my background is in journalism. That's where I started out, and through a series of weird circumstances that were happening in the recession of 1991, that's so far ago most people probably won't even remember it or... And some of them weren't even born yet then, but I wound up working in business-to-business media and started out on robotics world and modern paint and coatings. And since then I've stayed in, and I started off with, on the discrete and processing sides of covering those technologies in those industries and have never left it. So it's been an interesting 30-year ride to see how this developed over time.
Don: Wow. You remind me from our conversations, 'cause I don't think it's a decade yet, but it's getting close to that, that we've known each other. And I too do not have the automation industrial background, I came to this from journalism also, as we've had conversations over a beer once or twice about our journalistic backgrounds. And it was as a group publisher with a group of technology magazines in the public sector. I didn't really know much about automation, except a little bit of crossover when it came to maybe utilities and special districts and water, waste to water public works kind of stuff, but not the industrial sector, not discrete process manufacturing, anything like that. So we both come from the non-technology side, but your robotics intro that I didn't know about, it actually opens up a part of our discussion which we'll get to later about robotics, so I don't know if they've come full circle, let's say, they've matured and evolved, and it's a bigger part of the discussion today, I think.
David: The one thing I can recall when I started on robotics were the editors who were the age I am now, and I was right out of school, they were all telling me how, "Oh, you missed it with robotics, 'cause you weren't here in the late '70s to the mid-80s, the whole robotics run-up is done." I've missed the revolution in robotics by not being in there when all the automotive manufacturers were buying in. It was sad that I wasn't there for the good old days.
Don: The good old days.
David: They were already the good old days by the early '90s. And it's kind of funny in retrospect that that wasn't even scratching the surface with what's happened with robotics since then, but the industry was saying, well... It's that it's similar to that idea that all the computers that had been sold have been sold... That would be sold, had been sold, that idea with no idea that what was to come. It was kind of similar with robotics. Yeah.
Don: Yeah, it's funny, if you go way back to IBM and Watson, I remember putting slides in my presentation years ago that the prediction was there was a worldwide market for maybe four or five computers. You know what I mean? It was like, if we can get four or five of these out there that would pretty much cover what the world needs, you know.
David: Yes, that's the main thing.
Don: You think about the evolution. Or even more recently, when Steve Hechtman, our founder and CEO of Inductive Automation, I can remember going to the IAS conferences in 2007, probably, and we were just getting started as a company and talking about... We didn't even have the Ignition platform, just legacy products promoting, and people would come up and say, some other competitors, "Why are you starting a SCADA product? That's a totally mature market, there's no opportunity in SCADA. It's wrapped up." So we heard the same thing, David.
David: What more can you do with this? It's been done.
Don: It's already been done, exactly. And here we are in the world of digital transformation and robotics again, and there's a lot still to do, that's for sure.
David: Yes, absolutely.
Don: Well, listen, let's get into that then, because you mentioned in your background something that is one of the reasons we really wanted you as a guest, which is you cover a lot of stuff, you've been covering a long time, and you do have the view of a journalist who's now gained expertise across the entire industrial sector. So I'm gonna open the door for maybe just a first question, just 'cause I'd like to get just an overview and then we'll dig in, but if you were just saying, what are the big deals? What are the big issues? The kinds of things we might wanna talk about, you just take a minute and say, "Here's the big picture that I see if I look at the landscape that we're facing right now."
David: I think coming at it from my history and what I've seen in the industry covering it, in my first 10 years in this industry were very much focused on the enterprise level software technologies in manufacturing, but it was largely around that. And so it's been interesting to see that where we've come now with everything around the Industrial Internet of Things, Industry 4.0, all of this that's emerged in the past 10 years, it's not a repeat of that, but it's more of an extension of all of the things that I heard early on in my career or what I saw evolving with how MRP became ERP. And a lot of the same things that happened with that outgrowth of a very simple technology to include more and more parts of the business as a whole, we're seeing the same things now. Except now, as opposed of going across all of these enterprise-level aspects it's getting granular down into the plant floor, the sensor and device level, and feeding up into it as opposed to corralling all of these enterprise level aspects, but it's a similar process. So I guess with my perspective, I've seen one thing and that it's like what started in the enterprise space has now come full circle to incorporate everything across the space down to the most granular levels of automation.
Don: Well, if you think about MRP, your materials resource planning, and as you said, it's grown and expanded. Now you've got ERP, and we're talking about the entire enterprise. Obviously, that also speaks to going from sensor to cloud, 'cause you're going from plant floor up to the cloud, all the technologies that empower the MRP and evolution to ERP. Is materials resource planning still used today? How does it fit into this larger ERP landscape that you're referring to there?
David: Yeah, when I came into the industry in the early '90s, that was just when MRP 2, which is manufacturing resources planning with the extra additions, there were actually some... I remember some instances, it was an editorial question of would this be called MRP 3 or would it be ERP? And that's about the time Gartner... I think it was Gartner, I think they first coined the term “enterprise resource planning,” which is essentially MRP 3. You basically had the early MRP, which is material requirements planning software. It got big in the '70s. I think it started in the '60s, but got big in the '70s, and it basically just is very manufacturing-centric. It focused on controlling the level of materials that you needed to produce whatever it was you're producing. It answered questions like, what items do you need for production? When are they required? How many are required? But then MRP 2, the Manufacturing Resource Planning, that came about in the '80s, and it was really an extension of MRP.
David: It went beyond just what you need to produce X, but it also looked into the master production scheduling, the bills of material, inventory control, purchasing management, and even into sales and forecasting. It just basically started connecting all the other dots that are connected to MRP anyways, it just started physically connecting inputs from those other systems. And then where it grew into ERP, that where you got into financial and accounting management, customer relationship management data, human resources software. All of that was put in so that you had this, a single consolidated database and a single screen too, for which you could access all of this information to see how when orders were input, what would be needed to produce and to give you information that was so you could do capable to promise and capable to deliver. So it was just kind of this continuous adding-on of capabilities into a single system. And I think to your point, what you were saying about things being aggregated into the cloud or into an enterprise system, that's kind of what started there has just kept growing and growing to include more sources of input and data for greater visibility, decision-making and things of that nature.
Don: Well, when you think about that, that evolution from MRP to ERP, and it seems like it really has continued to just grow and grow and grow. What do we want in terms of data and what do we want to be able to do with it. Do you see that as setting the stage for the evolution of OT/IT convergence and one enterprise, and not any limitations on what you can do with your data and stuff, is this part of an evolution that we're then continuing to see unfold around us every day?
David: I do, I do. I see it as exactly that. I think what started there is just a natural outgrowth around that, it's like... That you start with MRP and you realize if you had more data from this system or that system, you could be more accurate and have more insights, and that just has continued to grow. And now we're down to the sensor level data, incorporating that into the analysis and decision making and the very... Even into artificial intelligence-driven analytics, we're seeing with that. So I guess having been here and seen it progress, it's almost an... I don't know if you'd call it a natural evolution, it has been driven by distinct business decisions and industry decisions, but it seems to make sense. When you look back on it, it doesn't seem that crazy. It seems like, well, you started here, you added this then you added that, the next logical steps are to keep adding to it so that you get this whole complete picture of operations with nothing missing, ideally.
Don: Ideally, yeah. And when you think about IIoT, the Industrial Internet of Things and what that brings to manufacturers, I can't help but think that... Every day we still see customers that are out there with whiteboards and clipboards and Excel spreadsheets, and it amazes to me that you see not only legacy equipment in the brownfield world. ‘If it ain't broke, don't fix it’ kind of mentality, and that's gonna be around for a long time. In some industries, you got 30-, 40-year-old equipment, but you still need that data, you still need to bring in that data, so manufacturers have to see value in order to actually take, in some cases a first step, in some cases many more steps on this evolution. So can you talk a little bit about what does IIoT bring now, and how important is it for manufacturing organizations to adopt it and get on the path?
David: Covering industry from my side and from your side as well, you know, producing the technologies that manufacturers uses, the one thing that we all try and stay conscious of is that manufacturers don't go into business to buy technology. They go into business to produce and sell X and make their profit and do what they do. The technology is merely a tool to help them do it better, reduce costs, improve uptimes, and so that's where the value that IIoT brings, it’s... When you look at it, going back again, we keep going back to this, I guess, 'cause it makes sense, if you're gonna talk about where we're going or where we're at, you need to look at where we've been. Even with ERP manufacturing, one of the complaints about ERP systems back in the day was that it treated manufacturing as a black box. All of this information and data was in there, but there wasn't a lot of visibility into what data might be missing, how the decisions were being made, was it based on incomplete data, and that depended on what sort of functionality was in the type of ERP you bought and how many connections were made to it, so manufacturing was something, it was like...
David: There was that component of manufacturing data within the ERP that was essential to its operation, but it wasn't very granular, it wasn't necessarily detailed, it could vary from product to product, but with IoT, the value of that is it... The whole purpose of that is not just what a particular supplier's product does and what it connects to, it's more of the concept of gathering all of the data that you can from your devices and systems that are in a line or in a production facility, and using all of that data to make decisions on. And I think that's the value of it, is that manufacturing is no longer a black box, so to speak, with IIoT. It's, now you know what it is and you can see where gaps might be. If you're going at it from a predictive maintenance aspect, if you will, and you're looking at that, and you don't have the same types of sensors collecting vibration, temperature and whatever might be on every one of your devices, then you've got an incomplete picture. And the IIoT approach highlights that and shows you that. Whereas ERP was like, it's all in there somewhere, hopefully. Whereas this you can see, and I think that's the value it brings. It changes the thinking about approaching the specific granular aspects of manufacturing data.
Don: Sure. You know what, you made me think about the fact that Arlen Nipper, one of our strategic partners with Cirrus Link Solutions is... Had conversations with me over the years when I was first getting a sense about this. And when we first started using MQTT modules with Ignition to get to the field, to get out into the field, he would take a look at something. I remember doing demos at conference and stuff, he'd go, "You know, you had three or four data points that you were getting from this particular pump or valve or whatever out there, but anything else you gotta come out, you gotta look at it. You don't know anything about the RPMs, about the life of that piece of equipment, how long it's gonna last or not last. And all of a sudden you can bring in 160, 250, whatever data points you want." But if you want business, you want... Big data analytics, you gotta have big data access, and that means you have to have, to your point, all of those data points out there. But here's something else I wanna build on, which is manufacturers are not going out there trying to find technology to buy.
Don: Being a vendor, I understand this. Vendors sometimes get really cool piece of technology and they... They're trying to sell a piece of technology looking for a problem, sometimes. The manufacturer just wants to run a good operation, so when you think of certain technologies like let's take the evolution of Ethernet is you got... You got a multi-decade history here of evolution, what aspects of Ethernet made it a suitable choice for plant floors, and now as we go to wireless, is wireless networking, a viable option for the manufacturing plant floor as we evolve?
David: I think with Ethernet it's an interesting one, 'cause it's clearly viable and is being used, but I don't think that's what drove it to begin with. Again, getting back to ERP, some of the first instances that I can recall of hearing of manufacturers running Ethernet down to the plant floor was to connect the ERP systems to their MES systems to get more of the granular levels of data that were available at that time before the ubiquitous use of sensors for IIoT that we're seeing now. MES was kind of the grand holder of all of this manufacturing execution information and tracking data from the actual production processes, so to give that greater levels of granularity to an ERP system so that manufacturing wasn't such a black box, so to speak. That's where the first instances of Ethernet coming down to the plant floor that I encountered occurred around. So it was more driven top-down; the ERP systems, the owners, the managers, the plant managers, the C-Suite wanted that data, so that the ERP could be more exact. So it was more top-down driven, from my perspective, as I saw than it was... Nobody on the plant floor was clamoring for Ethernet.
David: As a matter of fact, I probably have told you this story before, it was a very illuminating one for me, this is probably in 2003, 2004. So about almost 20 years ago now, I was at an industry event and it was the night before the event started, there is the welcoming reception, the typical reception that you have at various industry events. And I was talking to some engineers at the event over a couple of beers, and I asked them what they thought about what at the time was an increasing movement to gather this MES data and feed it into ERP systems. I asked if they were doing that, were they looking at it? And this guy, he says to me, and he comes into my face, puts his finger in my face, pointing at me very close, he goes, "Ethernet is not a viable industrial networking technology." And he went on to just go at it about how... Mostly for its lack of determinism, but just on and on. He goes, "That's a front-office technology, it's not plant-floor technology." And he ended up saying that Ethernet will come on to his plant floor over his dead body.
David: And that was the state not even 20 years ago, that was not uncommon to hear. So that's why it was really a top-down-driven thing, which now, with what's going on with Ethernet, from time-sensitive networking, which brings determinism to standard Ethernet, to single pair Ethernet, and Ethernet-APL. The things that we're seeing, the advances around that, those sorts of changes have really made Ethernet a plant-floor-viable technology, where... It was capable, but now, it's got the extra things are being added to its capabilities to really make it a true plant floor network and not a network that can work there, but it might not be ideal.
Don: Well, interesting that you say that from Ethernet, but not that many years ago, and having been out to oil and gas customers and talking, I have another conversation I had with Arlen Nipper about an oil and gas customer who said to them, "The cloud will never touch our networks."
Don: It will never see an industrial organization. And probably was wagging a finger in someone's face, just like at you over Ethernet!
David: It's the same thing, yeah.
Don: I know, it's the same thing. I guess you also realize that sometimes the biggest challenge we have with the evolution of technology and misutilization is not the technology, it's we as people.
Don: Change doesn't come easily, and to see the art of the possible sometimes takes a little bit of a stretch. Let's go into another technology, though. I gotta say mobile. So, you have Ethernet, but now we've got mobile devices. Have they changed the way we approach industrial automation? Is there some big impact where you can say that panel PCs and HMIs are gonna be obsoleted by mobile, or how do you see mobile fitting into the landscape of the evolution of industrial automation?
David: I think it's still actively changing it. It has changed it, and we'll talk about that, but it... I don't think that is nearly done yet. I think what we saw happen with COVID, which forced a lotta people to be... Open up to remote access to systems has changed a lotta things as well, and that's going to facilitate the greater use of mobile devices in plants where that used to be more restricted. And there was the whole discussion years ago about BYOD, bring your own devices; what companies allow that, what companies mandated use of their own mobile equipment once you were in the plant, but... Yeah, it's changed it in terms of the actual plant floor workers and as well as supervisors, just the ubiquitous ability to access information of all types at all times.
David: And the main areas we're seeing this round are obviously in maintenance. The maintenance personnel can come into equipment. They don't have to plug in the laptop. I can't remember seeing... I know they would actually have to physically plug their laptops into machines to get data from it. I'm sure it still happens, but that's largely unnecessary now with mobile devices. And training as well. We're seeing that with augmented reality through the mobile devices. You just hold up a tablet or a smartphone to a device, and it can pick up the QR codes or other ways of doing that to bring the data to show you to diagnose what's wrong with the system, to access material safety data sheets and other operating equipment. So yeah, I think mobile is in the process right now of changing the way operators do manufacturing.
Don: I was gonna say, the mobile world also is one in which sometimes you don't know what you don't know, so when you didn't have mobile, you didn't necessarily think with all the possibilities. I know with solutions in there... The industry, you cover all aspects of vendors and what they're doing in the industry. So solutions like Ignition Perspective, those have opened up new possibilities in industrial SCADA. And when you think about what we see proliferating around Perspective as it's matured... I know when Carl and his team in development were working on it, it was like... When he finally released Perspective, he said, "This is the mobile solution that I always wanted to develop, because it really allowed one to take advantage of mobile technologies." So when you see that kind of evolution out in the marketplace, what do you see as the potentials for solutions like Perspective that are really gonna change the way people think about the possibilities?
David: The big thing with a technology like Perspective is that it's... I don't know if I'd say changed, but it's definitely expanded how you think of HMI/SCADA, 'cause HMI/SCADA, when you think of it historically, it's been however the integrator or the technology vendor set up your machine based on your inputs, of course. Your screen was your screen; what you saw is what you saw, and that was it. And if you wanted to change that at any point, you had to call the integrator or technology supplier back out, you had to do changes, and that could be very costly. But with Perspective, it's basically made the design... Because all of the data from the systems is coming into the software that you can create the screens you want, so it shows what you want to see, how you want to see it, and if your needs or responsibilities change over time, you can change those screens that you see on your device based on user access or permission to see certain things within that. So I think that flexibility within the software to be able to change on the fly and not have to... To do it for the end user to be able to make those changes themselves.
David: And then as well as to the other aspect of it is how it brings the capabilities of the device itself, the smartphone or the tablet, into the capabilities of what HMI/SCADA can do: The GPS locator within the phones, you can use that. Use the camera as a barcode scanner with that camera, things like that, but it's... So it's not just the screen showing you static data or a static screen feeding from these various sources. It's a piece of software that can change as you need it and use other technologies that are inherent to the device that is being used on to affect what you're seeing.
Don: Yeah, the... I think back to, as you talk about all that flexibility, I think back to Mike Milinkovich, the Executive Director of the Eclipse Foundation a couple of years ago at our Ignition Community Conference, and remember the old days when we used to have live conferences in person?
Don: We're going to again. It's coming back, it's coming back.
David: Looking forward to that. It's been a while since I've been out there, yeah?
Don: Yeah, it's a total digression. My wife said to say hello to you, because I remember us at ARC, sitting there, and she's... The man plays guitar in his spare time and has all these other talents and interests and stuff like that, but we've... We so enjoyed that ARC conference.
David: Tell Dee I said hey back.
Don: Yeah, I will definitely do that. But we sat and had some sushi and a couple of beers and enjoyed an evening, you know? So I... Back to the point, I was gonna say, with all this kinda about software development, Mike Milinkovich made the comment, "Software is eating the world."
Don: So, with that, you also cover lots of hardware, lots of people building stuff. So with the introduction of mobile devices and all this software capability... Products, like the Perspective Module, making remote access with the push that's happening, that hit us from the pandemic and really pushed us into the remote world faster, I think. How does this affect the development and sales of OEM equipment? How is this gonna impact on the hardware world, from your perspective?
David: Just getting back to the software aspect, we've all heard of Software as a Service: The software you subscribe to, you don't have to... You don't have all the CDs or discs anymore to... Or even to need to download it on your computer to install it anymore. You just subscribe to it and you get it. For Ajin, you're accessing it in the cloud of course, but you have the full access to the software. I think how this remote access... The use of, greater use and expansion of remote access technologies is going to contribute to us seeing Machines as a Service as well. It's gonna create a new business model for OEMs. And we're already saying, "This isn't some pie in the sky. It's gonna come 20 years from now." It's been happening for years now. Rolls Royce jet engines is one of the first. Now, of course, this is selling to aircraft manufacturers, who sell... Or they sell them to the airlines as part of their agreement with that to be installed on the aircraft that they operate. But they started doing that using Microsoft technologies to...
David: Basically, you don't buy, these airlines don't buy the engines anymore. They are buying flying time through that, and that's what Rolls Royce is guaranteeing them, the flying time. And the way they do that is the airlines are paying, I assume it's an annual base. I don't know if it's monthly or annually. I don't know how that's set up, but they are paying for Rolls Royce to constantly maintain and guarantee the operating parameters of the engines on their planes. So the... Rolls Royce takes us on completely; all of the maintenance, that they track where the planes are, they can tell them if there is a problem. They can see where the closest facility is with the part that they'll need. Like if they can tell like, "Okay, this engine's running fine now, but we're detecting that within the next 100 hours of flight time, this is going to start creating issues that'll need to be fixed." They can see that well before it becomes a problem, and they can see where that plane is going. And if it's on... Within the next 10 hours of flying time it's gonna be in an airport that has that part, they'll go ahead and replace it then, long before it ever becomes a problem. That's part of the Machines as a Service, if you will. I said they're not buying these engines and maintaining them themselves, the company's doing that.
David: And I see that happening in the OEM space across the board. One of the ones we've been, as part of PMMI, the packaging machinery... The PMMI media group that we're a part of, that we see that happening with even packaging machines potentially. I don't know of any examples yet, but it would be the same thing, where a manufacturer who's packaging their end product, right now, they buy those machines, install them, and then have to have separate service agreements with the OEM or a third party. Whereas in this case, the manufacturer would just contract with the OEM to guarantee that they can package up so many packages of potato chips or so many bottles of beer, whatever their production parameters are. And that will be guaranteed by the OEM, who uses remote access technologies to constantly diagnose and maintain the machines and keep up with them, so that the machines never go down. It kinda changes that business model from buying the machine to really buying production capability. And so that's big potentials for changing the business dynamics of how manufacturing is done.
Don: Yeah, it seems like... So you have Software as a Service, we certainly know of Platform as a Service. We're actually going all the way to you're gonna get this piece of Machinery as a Service, so are we... Are we seeing an overall trend from your broad view there to embed control and communication functionality right into the machine as we move forward? So you're basically getting functionality, communications, control technology, everything you need at that machine level, to basically buy the number of potato chips you can package. Or you're buying that functionality and that production capability, and you never own a machine.
David: I've seen... [chuckle] Having gone through, as we were discussing earlier with the whole Ethernet, it was a crazy idea to have that come onto the shop floor. I don't see this... This is a bigger change, a much broader change, in an intellectual capacity change about how you think about the business. But in a way, it makes more sense. Getting back to what we were saying kinda earlier is manufacturers aren't buying technology for technology's sake. They're buying it to produce X amount and to keep the business running and expanding, and this kinda gets to the heart of that. You're paying for this service to guarantee your ability to produce X. Not buying machine and hopefully, we can keep it running to do X, which is kind of where we're at and where we've been. Now, it's a whole different concept, the big questions is how successful can OEMS be at this long-term? This is still... Even though, like I said, it's not brand new, it's been happening for a few years now, with the Rolls Royce example and some others, is... As, I guess, we'll see this over time, over a decade, how successful this is. And if it is that, well, I don't see why it wouldn't change the dynamic of manufacturing and technology use.
Don: Yeah, it's a long road from the folks whose mobile device is a clipboard right now.
Don: To the kinds of evolution that we're talking about here. And so, if you think about this going to the top, and to the cloud, and... The technologies: AI, machine learning, the whole world of industrial controls, and predictive. You mentioned predictive analytics and preventative. So, what do you see, AI and machine learning, how do those fit into this digital transformation evolution, this IIoT evolution we're engaged in?
David: On one hand, I think it's a fantastic tool, and it's a tool that's going to be used and needs to be used, because there's no way the amount of data that's being collected now by all of these sensors that we've been talking about and all... Connecting all of these systems to feed data into a single database to produce the single panes of view where you can analyze all those data, it can't be done by humans. And it's gonna have to be done by artificial intelligence and subsets of it, and machine learning and stuff. And we're seeing a lot of it, like you mentioned with predictive maintenance, but also in machine vision and quality as well. We're seeing that used a lot. My concern about this is... It was actually... I have now since been on two trips since just in the past couple of months, gone back into traveling, but my last trip right before the pandemic hit and kinda shut everything down for travel, was to go to a TechCrunch event in Berkeley, and I saw... One of the speakers there was Stuart Russell, and he wrote that book, Human Compatible: Artificial Intelligence and the Problem of Control.
David: And he was basically saying the issue he sees with AI... He's an advocate of AI. I should start out and say that he's not a detractor or a Luddite when it comes to that by any means. But he says, "We're thinking about AI the wrong way." And the reason why is he said the standard model of AI that's broadly used, not just in the industry, but across the board, is one that was borrowed from machine intelligence back in the 40s and 50s. And he said, "In the... That idea is that AI is a rational agent that's been created to achieve an objective." But he said the problem with this is machines don't set objectives. Humans have to plug those objectives in, and that's where the problem is. And he... And his too. He gave a couple of examples about why that's a problem for humans to correctly set an objective. He talked about the rubbing the lamp and the genie, and you get your three wishes. And usually, the third wish is to undo the first two wishes, because you didn't foresee the outcomes of that. And then, the King Midas thing, where everything he touched would turn to gold, and he wound up nearly starving to death, 'cause when he touched his food, it turned to gold.
David: Not being able to see all the ramifications of the decisions we make, that's what he sees as the problem with AI. So he was saying, to counteract that, what we need to do with AI and be more conscious of, and I'm not... The one concern I have is, I'm not seeing, necessarily, a lot of effort being put into AI, at this area yet, is that AI needs to be constantly beholden to humans. It's kind of like Asimov's law of robotics, that a robot would never harm a human or would allow a human to be harmed no matter what. It's the same thing with AI. He says a machine can't be allowed to assume that it has the complete objective. It has to know... The machine has to know that it doesn't have the whole objective, and therefore, has to be obligated to optimize for human preferences, so that, as the machine solves problems, the better the result will be for humans, not necessarily worse.
David: But I think that's something we're not thinking of. I think it's gonna be huge in that, but I... That's my one little concern with this. And I mean, at a machine level, in the industries we're talking about, that's not something day-to-day worry about. But it's something about how the construct of AI, 'cause as we've seen, going back to the MRP, the ERP thing, everything is built on what comes before, even up to IoT, is really just an extension of the concept that... Around ERP. So that's the potential worry if we just keep expanding and tuning with the same format in that and don't correct that soon, that could be potentially problematic. One day, when that singularity of AI occurs, when and if that... Now we're getting into potential science fiction, but hey, it's like, people would look at us today 50 years ago and think it's pretty sci-fi, so who knows?
Don: That's exactly right, and one of the great things about podcasts like this is we can talk about whatever we wanna talk about, and I think those are important questions that need to be asked. The human element and the maturing of the human side of the evolution of AI is, I think, a critical component. So now, let's circle back to your robotics. You entered this industry in the robotics world. So, with Industry 4.0, IoT, all this evolution we're talking about, how close are we to having robotic collaboration? So, obviously, we're going in a direction where we could have robots collaborating with robots, and the evolution of that means... Not that the human doesn't have a role to play in it, but there's a lot of independence that starts evolving when you get technologies like robotics going into that level, so comments on robotic collaboration and its evolution?
David: Yeah. Well, we're already seeing, at an early level, with collaborative robots that can be programmed to work alongside humans, and if there's any sort of interaction or it gets too close, the robot slows down. The whole idea of speed and separation monitoring, as well as the touch on the actual... Many of the collaborative robots, if you touch it in any way or if it touches you, it will slow down, not cause any sort of injury. And they don't move very quickly either, most collaborative robots, so even if it did hit you full force on, as long as it didn't have a sharp object, injuries are gonna be minimal. So we're already seeing it to a good... Bit less here, but I think a lot more with the evolution of AI, we're gonna see greater levels of that. And again, it gets back to what we're talking about, hopefully, that will be done correctly and right, and some of these issues will be addressed with that. In terms of what it means for the industry overall, you have this class of collaborative robots now that we're seeing more and more of them becoming familiar with, but this next step I'm seeing is where they're taking industrial robots that have the higher payload capacities and move at higher speeds, to where they're introducing collaborative capabilities with those robots, through speed and separation monitoring and other 3D technologies, that look at the entire potential work envelope for a particular robot.
David: And there's... So that you're not buying a special collaborative robot anymore, but you can buy an industrial robot that can do all of the things those robots can do, but have collaborative capabilities built in, so that they're safe to operate around in an uncaged environment. This is still relatively new, but it's being used. I've seen it from a couple of different companies now that are introducing this. I guess the big tell on this will be what level of uptake it gets across the industry, and how... It seems like it's got a lot of potential upside. To be able to operate... I mean, some of these massive industrial robots will probably always work in a caged environment or behind special safety systems for that. So I don't know how far it will go in the near term, but for smaller industrial robots, high-speed pick and place, things like that, that's this... Collaborative capability's already come, where they... By the time human gets within five or six feet, it starts slowing down dramatically. And by the time a human gets within two feet of the work envelope, it completely stops. And as the human backs away, it starts to slowly restart. You don't have to restart the system, it restarts itself. That is kind of the ultimate, at this point, of what we're looking at in terms of robot collaboratively.
Don: I appreciate your comments on that, 'cause I think it's a fascinating field to see that evolution, and what will be the uptake? That's, of course, the open question as we see it evolving because you saw it with Ethernet, we saw it with industrial information going to the cloud. And now, we're seeing it in all sorts, from mobile to robotics and robotic collaboration. So we covered a lot of ground here in the last 30, 40 minutes, whatever it is. I just... As we're sorta coming to a close, I kinda wanna just ask you an open-ended question about anything else on your mind. Anything else you think you wanna share to our audience here?
David: I think if anything, just having seen how industry has come through these changes, and I've, as a journalist, just been following along and saying, I think if there's anything that... And I still see it in some and manufacturing, and it makes sense, it's human nature, is to shut down the idea of new things too quickly, sometimes. If anything, I think anything in the past 30 years, and you could probably go back further, and that has shown us is, to really be open-minded about this, because some of these things that are very easy to dismiss now won't be so easy to dismiss in the next, not 10 years, but two or three to five years. So I think that's the key thing, is to keep an open mind about this while not... While, of course, staying grounded in what the needs are and requirements are. Not going full technology for technology's sake, but being open-minded to the possibilities.
Don: That's, I think, a really good... That's a good little final statement as we wrap up here. I just wanna comment on it. I think there's always a challenge with any newer technology and any organizational evolution, that you have to balance velocity and stability. There's some balance of, "Yes, let's go fast," but we're not trying to get bleeding-edge technology that's gonna destroy our stability, particularly in manufacturing. Back to the original point you made, the manufacturer is not looking for a technology, they're looking to build a better organization; more efficiency, handle downtime, those things. They're trying to get a product out the door. And that's gonna be the focus that we in our community, have to keep in mind when we're serving those customers. I just wanna say overall it's always been great over the years, working with you and seeing your perspective on things, and you've got a great opportunity, as your role, to look broadly at all areas of technology and have some opinions, some editorializing, and some observations about those. And so real pleased that you were able to take some time and share it with our audience today, so...
David: It's... It's been a wide-ranging discussion, for sure. So yeah, thanks for having me on. It's been great not only catching up with you, but seeing you and catching up with you through our link too. It's been, like you said, it's been a while since we've seen each other. And hopefully, that's gonna change here in the next couple of months.
Don: That's fantastic. Well, listen, I wanna say to you, your family, have a great holiday season coming up. And I'm gonna say that I'll buy you a beer in Orlando in February. Let's just put that out there on the calendar, and we'll make it happen and have a good finish to the year. And thanks again so much, David, for joining us today.
David: Thank you. And same to you, Don.
Want to stay up-to-date with us?
Sign up for our weekly News Feed.