Maurice Liddell: 0:03

There's a tremendous ROI on it and there's been several case studies on this to where organizations have deployed this, the wearables, the training and utilizing the augmented realities. The return on investment is, is magnitudes greater than what the cost is. And in addition, it's also helping with organizational reputation. You're starting to retain more of your workers because people feel safe. And no one wants to work in a unsafe environment

Tom Raftery: 0:32

Good morning, good afternoon, or good evening, wherever you are in the world. This is the Digital Supply Chain podcast, the number one podcast focusing on the digitization of supply chain, and I'm your host, Tom Raftery. Hi everyone and welcome to episode 346 of the Digital Supply Chain podcast. My name is Tom Raftery and I'm excited to be here with you today sharing the latest insights and trends in supply chain. Before we kick off today's show I want to take a moment to express my gratitude to all of our amazing supporters. Your support has been instrumental in keeping this podcast going and I'm truly grateful for each and every one of you. If you're not already a supporter I'd like to encourage you to consider joining our community of like minded individuals who are passionate about supply chain. Supporting the podcast is easy and affordable with options starting as low as just three euros or dollars a month. That's less than the cost of a cup of coffee and your support will make a huge difference in keeping this show going strong. To become a supporter simply click on the support link in the show notes of this or any episode or visit tinyurl. com slash dscpod. Now without further ado I'd like to introduce my special guest today Maurice. Maurice, welcome to the podcast. Would you like to introduce yourself?

Maurice Liddell: 1:48

Great. Thank you, Tom. I'm Maurice Liddell. I am a principal with BDO Digital, which is a subsidiary of BDO. And we are the fifth largest CPA firm in the world. And I primarily focus on helping organizations in the marketplace to leverage technology for improving operations and managing and mitigating risk. And I'm talking to you from hot Houston, Texas.

Tom Raftery: 2:15

Okay, great. And I'm, I'm talking to you from a hot Seville in Spain. we're at about 40 degrees centigrade at the moment. That's about 110 Fahrenheit, if, if my numbers are about right. We've had several weeks, well, over a month, in fact, of, of of temperatures you know, at or around 40 degrees. It's been, it's been pretty insane, I gotta say. This summer has been mad, and unfortunately, It's not going to get any better, is it? I mean, I think we will, in 2033, look back on the kind of cool, balmy summer we had in 2023, if things are, things are going the same way they have been for the last couple of decades.

Maurice Liddell: 2:55

Yeah. Well, Tom, so no, I'm not going to complain about my 38 degrees here.

Tom Raftery: 2:59

Okay. So, Maurice, we've, we've, we've seen, we'll move off the weather topic, we've seen huge, huge shifts in technology in the last, 6, 8, 10 months, I wanna say, you know, particularly the rise of AI has come really to the forefront of people's minds, particularly since the release, I will say, of ChatGPT in the end of November last year. So things have really kicked off. And it's, it's making a huge impression in supply chain, the same as every other industry. What are you seeing, seeing as you are, you know, deep in the technology and supply chain space? What kind of changes are you starting to see?

Maurice Liddell: 3:49

Yeah. So, so Tom, I agree with you. You know, it's been a topic of conversation with almost every you know, company that I'm talking to these days. Right. And so, you know, with anything new, there's always that, that FUD that comes with it, the fear and certainty doubt and, and also fear of missing out. Right. Am I, am I missing an opportunity here? And this is where we have really had a lot of great conversations saying, How are we leveraging this technology to really improve things, and in particular about, like, safety? And there, there's been a lot of advances in safety technology in the past, where we're taking a look at just, like, say, passive technology, barriers, color coding, and so forth. But we're really seeing the technology being able to help, I think, kind of elevate safety awareness and improving safety programs. And, and the way that I like to look at this is really it's a, it's a system. Right? And if you take a look at, first of all, from like a, a cultural perspective how an organization really embraces safety. Then you take a look at, of course, the environment and talk about like the, the passive barriers and, and how we, we protect workers. But then there's also get into elements such as the employee themselves and their training and, and then the processes that they have to follow. So we're seeing this technology being able to really affect every one of those, right? So, so if we start to think about, first of all, the culture and how can we, like I said, leverage technology to improve the culture we start to think about how does that affect profits, profitability the overall, I'd say organizations reputation. And so that helps drive establishing a good culture around that. But then we take a look from the employee and starting to leverage technologies like AI for helping with the training. Take a look at immersive technologies and being able to use virtual reality and augmented reality augmented reality to train an individual on a new on their processes and the environment before they actually get in it. That's helping to make a more aware, individual. And then we start taking a look at the overall, like I said, the environment piece of it. And this is where where it can leverage technologies like computer vision and being able to identify, potentially hazardous situations before they actually occur. So, so now we've really moved from being in the reactionary to, to being proactive and, and predictive based upon situations. And this is where I think it's some pretty exciting uses of the technology, like I said computer vision tied with artificial intelligence. What if we're able to see spills or debris in a certain pathway before there's any equipment or workers going down that path? Being able to alert to that and being able to resolve it. Or, what about if we can leverage it to identify these safety zones? And so if you see movement within a certain type of safety zone, that would be a big red flag and we'd be able to alert and take some preventative measures, you know, around that. So these are all exciting conversations we're having with organizations and getting deployed, these types of things. It's not, you know, this isn't future. Right. This is actually today now. And ChatGPT is, like I said, conversation and a topic, which is helping to drive those conversations.

Tom Raftery: 7:33

Fascinating. Fascinating. I assume it could do things as well. I mean, you mentioned safety zones, but for example, I assume it could spot if a worker wasn't wearing a helmet or, or, you know, other safety equipment as well and alert to that.

Maurice Liddell: 7:49

That's absolutely correct. So, we're actually utilizing these technologies today for those exact things. So you can define all sorts of different types of rules of which the, the technology can help to identify and implement. So are they wearing the proper PPE or are they in a location where they're not supposed to be, or you see the convergence of man and machine forklifts and workers. So that, that's a very common type of incident is where you have human and machine interaction from material movements and being able to just identify that and, and make alerts, either visual or enunciations can help. And then in addition to we're seeing a lot of uptake in the use of wearables. And not in utilizing those wearables to actually start with training and we've seen where, where organizations have built a program, they, they outfit a new employee with wearables and it's monitoring their work activities is how they exercise their duties and being able to identify, you know, times when they're under stress. Or if they're getting exposed to certain types of materials, gases, so forth. And they can use that information then to modify their, their work procedures and processes. Right. And so I said, there's a lot of different, I think, fascinating applications of this technology.

Tom Raftery: 9:16

Yeah, even, even my own watch, put on the screen there, even my own watch, just an Apple watch can alert me to things like when I'm in an area with excessive noise. So it'll, it'll send an alert saying, you know, it's exceeded a certain number of decibels. You want to be careful. So I imagine that would be very useful as well in the, in the industrial zone. You know, if you're in a manufacturing plant or something, you want to make sure workers are wearing ear protection in those kinds of scenarios.

Maurice Liddell: 9:43

Absolutely. And so, you know, with this technology either it's a wearable or it's other type of environmental systems, those situations can all be identified and being able to help, like I said, be preventive versus reacting to when a situation does occur.

Tom Raftery: 10:02

Sure, sure. And it's not just about keeping workers safe, is it? Because if you have an accident, you can then potentially require downtime and investigation, you're losing time, which otherwise would be used to keep the plant running and keep productivity up.

Maurice Liddell: 10:23

No, that's absolutely correct, Tom. And so, we saw early adoption of some of these AI technologies to really focus on that, like I said, preventative or predictive maintenance and being able to identify when a particular piece of equipment may be going out of calibrations, either through vibrations or temperatures current draws. These are all things that can easily be monitored and that will give you a telltale sign that something's not right. And being able to kind of schedule and fix that because it's always cheaper, easier, faster to be able to provide a simple fix, versus potentially replacing a whole piece of equipment if a, a belt were to break, for example, and that can cause all sorts of damage, as well as create a safety related issue for, for workers. So those two definitely go hand in hand.

Tom Raftery: 11:19

Okay, and I mean, another use of the likes of these large language models has to be to facilitate maintenance because just a couple of examples from, from my own home here, for example I wanted to find out where a particular component was in my car. And so I went to Bing chat and I. I gave it the model of the vehicle that I had, and I said which part I was trying to locate, and it was able to put up a diagram. A schematic of my vehicle pointing out exactly where that component I was looking for was located in the car. And similarly, I had an issue with the dishwasher in the kitchen. It's not a particularly common dishwasher, but it's an old one from a good while back. And it's a model that's no longer being sold. So again, same kind of thing. I gave it the model number, the make and model number, and I told it what component I was looking for. And again, it was able to pull up a schematic. So I can imagine in those kind of scenarios, particularly if it's something that's no longer being manufactured and maybe the manufacturer has gone out of business, finding those details might be challenging, but for something like a large language model, very often it's trivial. You find it in seconds.

Maurice Liddell: 12:34

That's absolutely correct. Tom, I had a very similar situation very recently and trying to an old truck and needing to you know, add, add coolant to it. And so I, I didn't have any idea where that port was on that truck and, and so it did the exact same thing. But, but you, you bring up a really good point here and we start talking about augmented reality and leveraging this technology. So, as a person who's trying to do maintenance, being able to display the schematics or the information in a very simple format, right? So maybe I'm trying to execute this particular procedure. You can give me details, detailed steps, and highlighting the next element of it. And being able to present like a 3D model of it and highlighting exactly where they need to focus on and even being able to demonstrate the procedure to it. So that definitely I think helps from an overall efficiency standpoint. It helps with educating the workers or then doing the proper procedures. And then like I said, so we get the equipment back up and running as quickly as we can. And in a safely in a safe manner.

Tom Raftery: 13:41

Yeah, yeah, yeah. And as well, you can find, for example, manuals for, you know, bits of equipment that might have gone out of date or generate manuals for pieces of equipment that have gone out of date too, right?

Maurice Liddell: 13:53

So this is where you start taking a look at these language models and being able to utilize it. You mentioned ChatGPT earlier Being able to generate documentation around this is just a wonderful, wonderful opportunity to get all these manuals available. So you can within the prompts be able to enter all the details around the piece of machinery, what you're trying to do, and say, generate me the operating manual or the procedures for doing this. And it will generate this within seconds. I said, so when we start taking a look at what I was talking earlier, kind of like the whole environment or system, so starting with the training element of it. This is something that can be extremely helpful because, you know, a lot of this equipment lasts quite a while and very capital intensive to replace. So, you know, take a look at some of this machinery. It's going to be in a production facility for many years. Being able to have good documentation on that, that we can train the new workers on to whereas you know people have retired the person that actually put that in may no longer be available. This is a great opportunity for leveraging technology for very quickly generating manuals And for operating procedures to like I said for education purposes as well as for efficiency purposes

Tom Raftery: 15:17

Yeah, yeah, yeah. And of course, some of the equipment, the manuals that came with it back in the day might have been written in oblique English or by someone for whom English was a second or third language. And now you can just say, well, this is the piece of equipment. Are you familiar with it? Yes. Act as a safety engineer and write out procedures for using this safely or, you know, whatever it is you wanted to do.

Maurice Liddell: 15:42

Now, that is correct. And you can even go beyond that and say you can do what are the common issues, right, and identifications of those. So you can even go beyond what just the manual is, but based upon historical information, these are the things that I need to be thinking about or what my issues can be. What are my steps to be able to do the troubleshooting around with it. So those things that didn't exist back then, now you can generate those.

Tom Raftery: 16:10

Right. Fantastic. Fantastic. What about using AI with robots, Maurice? Because I know that's starting to happen now as well. And what kind of potential do you think that has?

Maurice Liddell: 16:23

You know, with that, that's an area where I think there's some potential challenges, right? And if we kind of take a look back at maybe what are challenges for adopting this technology. There, there's a couple of things that we need to think about. But one of them is trust. And there, there's already individuals that just don't really trust robots to begin with. Either because they feel they might be taking their jobs. Or there, there's been so many science fiction movies about how robots have really tried to take over. You know, you take a look at IRobot was, uh. was one of them. And there's, there's several like that. So like trust becomes an issue. I think that's something where we're going to have to, take steps to be able to kind of prove these things out and be able to help build, like I said, trust from individuals on that. We're also going to have to work with I'd say unions and other workers try, because you take a look at some concerns might be big brother, someone's watching over me and someone's going to be taking my job when we start adding this into the additional robot. So I think that's kind of like an overall, I think technology adoption and change program where we have to work with management, we have to work with the, the workers and the unions to be able to like I said, have them part of the decision making process, get people comfortable with that technology. And then one of the other challenges too with all these things is just what about the connectivity and networking, right? So some of these legacy plants they are not necessarily wired for that type of technology. So here we bring in like 5G technology being able to help with that overall connectivity between the different devices between the robots themselves. And so, you know, kind of, kind of back to your, your, your question there, Tom. I get excited about the use of these technologies. I think there's a great play for them, but we're going to have to deal with those adoption issues as we start getting more and more to where we have equipment kind of acting as a human, if you will.

Tom Raftery: 19:49

Right. But I mean, the advantages of robots are well known now. I mean, you know, they can, you can send a robot into a room where a human couldn't go because of temperatures or pressures or things like that. You can have it lift loads that humans couldn't do or humans couldn't do safely or at least, you know, pick up a box that a human would find hard to do and a robot, easy to do, or mechanize a production line with robots doing tasks that are just too boring for a human to do, you know, and so the advantages of robots in those kind of scenarios. And in warehousing as well, we're seeing a lot of robotics starting to enter and automate as well. Right?

Maurice Liddell: 20:30

No, completely agree in those types of scenarios. Tom, I couldn't agree with you more. Right. We, we've seen a lot of robots being used for material movements and being able to do you know, just in time type of material movements throughout plants. And you talk about like the, the heavy lifting, the other type of dangerous environments. Completely agree. Those, those are great applications of it. What, what I'm starting to think is like, what's next with these types of things is we're always trying to think, how can we improve these things? And as we start thinking about what can robots do better versus what can humans do better? Things like quality control, where you can quickly scan a component and checking for things that are out of compliance. A robot can do that much better, much faster than what a human can. But certain type of assemblies to where we still haven't mastered or the, the movement required for these detailed assemblies, a human can do that better, but we can supplement them with the, we'll talk about like the schematics or, or the drawing. So a, let's call it augmented reality assembly line to where it can help project. This is where you need to do something, or here is a particular zone you need to work with. That's where I've seen you get a great complementary of man, machine, and technology. So, as we progress type in leveraging this technology, I think there's just it's unlimited of what we can do. But there's always going to be that point to where a human can do certain things better than what the technology can't. And how do we best kind of merge those two type of capabilities?

Tom Raftery: 22:22

Yeah, yeah, yeah. And, you know, speaking of that, of that, of getting humans and robots working better together, I can see using the likes of large language models as a way to facilitate that because if you combine the large language models of the likes of the Chat GPTs with voice to text and text to voice technologies, you can then start having a conversation with a robot and ask it to do something and it reply to you in just natural language as opposed to having type in commands and all that kind of thing. Suddenly you're a human to robot interface goes away.

Maurice Liddell: 23:01

That, that's correct. And so we, we use that today in our households, right? And you talk about, you talk to your phone or you talk to you know, the device is saying, Hey Google, or Hey Alexa. And being able to get that type of type of response, that's going to increase efficiency. It's going to help reduce errors. But you know, there are still, this, this technology isn't perfect, right? And it's just even like, let's just say we use Alexa or or, or Google. Sometimes you get a response back that baffles you a little bit, right? And, and so the same thing can apply here. And it's going to progress. It's going to get better as we do it with these language models. But yeah, it makes things so much easier. And especially when someone is like doing maintenance you talked about earlier. And being able to say, okay, can you give me the procedures for doing this type of activity? And then if it can speak to you of doing the step, you can speed it up, slow it down, skip steps. It just makes it much more efficient for an individual. It makes it into a format that that individual is most comfortable with. Some people like to read, others they like to be talked to. All right, and then, you can even utilize, I'll talk about, you know, progressing into holograms of being able to display a hologram around this and do a rotating 3D model around that. So that is where I think there, there's a lot of promise around those things, Tom.

Tom Raftery: 24:29

Yeah, yeah, yeah, yeah, yeah. I mean, if I just think back, I, I've been doing podcasting off and on since 2005 is when I did my first podcast. And the technology has advanced so much in the last 18 years in this space. When I did it first, you know, I had to create my own RSS files for the podcast. Now all that's automated. Now I can take the audio file today from this recording. Drop it into an AI app and have it come out with a full transcription for me within a matter of minutes. And then I can start, there's a, there's a program I use called Descript for the editing. And I can start editing the transcription. And as I edit the transcription, it edits the audio file. So if I remove a sentence from the transcription, it removes the edit from the audio file. And if I'm working with the video files, It works exactly the same, so I can just start pulling out sentences here and there from the transcription. And it automatically does it on the audio and the video file. It's just, it's mind blowing stuff, you know. It makes the editing function so much easier to do. And it also gives me a full transcription automatically. You know, stuff that was impossible five years ago. Never mind 18 years ago when I, when I did my first podcast. So, it's just, thinking about the way that has happened where is AI going to take us next? What's, what's coming down the line, do you think?

Maurice Liddell: 26:03

So, if we kind of keep along the vein talking about like from, from safety perspective, we started thinking about really moving more from, I'd say, passive type of safety devices to, to active and that are based upon situations as well as individuals. Right. So, say for example, there's a while back there, there was a big safety incident at a foundry where an individual had fallen into a vat of molten metal. Right. So of course that, that was a very devastating situation. But they fell off a scaffolding. So slip and fall. Yeah. And that's what happened. What if we could augment that with, I'll call it dynamically deployed safety nets in certain environments. So individual wearing a wearable or do you got computer vision, see a person slipping and being able to deploy a net over that molten VAT. And so which would prevent the actual death. So I think we're going to continue to see an evolution of like safety devices as well as what you talked about these large language processing models. How are we going to leverage those for other types of features? It's going to get into conversations with machines. Now and you talk about you know robot a lot of people think about something that kind of looks like human or it's moving around. But it can actually be a piece of equipment on the floor that is doing a process. So we're used to seeing all these different screens for the man machine interface, Now we're talking to it, or we're giving it instructions via our voice and then it having enough intelligence that if we tell it to do something that may not be within the, the right procedures or that could be a safety related issue, it can alert us. to that fact So help us from, you know, harming ourselves, if you will.

Tom Raftery: 27:57

Yeah, yeah, yeah, yeah. I remember when ChatGPT first came out, one of the first things I tried doing with it was speaking to it, not typing into it, I should say, not speaking to it. I typed into it a message in Irish. Now, Irish is a language which is spoken by, you know, a very small number of people, and I didn't expect to get a reply, but sure enough, it replied to me in Irish. Now, what it said in its response was, yes, I can speak Irish, but not very well, so I might make some mistakes. Which I thought was fair enough, but I was impressed it could say that in Irish. But the point I'm getting to is in manufacturing, in warehousing, in lots of these environments we have people speaking different languages in the same building often. So again, this, these large language models, because they can speak not just English, but any language you throw at it pretty much, that's got to make as well the workers in these environments feel more comfortable. being able to speak natively in their own tongue to the devices they're going to be working with, right?

Maurice Liddell: 29:05

Absolutely. And I think, you know, another element to that, too, Tom, is not just from the particular language, but it's also, I'd say, style. And, you know, with that, we can definitely see a bias in these models, right, because they're trained by information that we provide them and so I think we have to just be conscious about that information that we're feeding it and making sure that we are not introducing our own biases into it and so that we feed it enough information that has I think a broad type of I'd say, just a broad type of data. Versus just from one type of source and and where we've seen this happen in the past is like when organizations use this for hiring practices and Saying okay, let's take the information on our most successful workers and maybe it's a resume, maybe it's their profiles or information based upon the performance reports, feed this into our models and now let's leverage this to identify, you know, from all the resumes that are out there, who are going to be the most likely next successful workers. Well, if the vast majority of those people that the analysis has been done on are males, you're going to start excluding females. Yeah. Right? So, so that's what I'm talking about. We, we need to be kind of conscious that we do have certain biases in these models and, and, We, we need to just think about that and make sure that we're not introducing those types of biases and compensate for that.

Tom Raftery: 30:42

Sure, sure, sure. That makes sense. Yep. Maurice, so we're coming towards the end of the podcast now. Is there any question I haven't asked that you wish I had or any aspect of this we haven't touched on that you think it's important for people to be aware of?

Maurice Liddell: 30:57

I think it's just... You know, one thing I'd like to just emphasize is that early on, I talked about kind of like the overall, it's a, it's a safety system and it's really based upon overall safety culture. And so I think we've all heard that, that statement of by, by certain companies saying people before profits, right? This is where I think safety can really help an organization maximize profits, you know, by, and you talked about it earlier, Tom, to where as if we do more of the predictive and preventative maintenance. We're able to provide workers with information about how best to do the procedures and we flag it if we think it's something that's not going to be safe. So I think that, that phrase really changes that, you know, they're, they're not mutually exclusive between safety and profits, but I think it's really thinking about safe workers help maximize profits. And so I think it just kind of changes that narrative a little bit because I, I know that you know, some organizations that we, we, it's not in the budget or, you know, we're really focused on our certain, core operations. But like I said, I challenge that and thinking into that by helping our workers to be safer through this technology. There's a tremendous ROI on it and there's been several case studies on this to where organizations have deployed this, the wearables, the training and utilizing the augmented realities. It's the, the, the return on investment is, is magnitudes greater than what the cost is. And in addition, it's also helping with organizational reputation. You're starting to retain more of your workers because people feel safe. And no one wants to work in a unsafe environment, so I said it just has a tremendous amount of benefits and the technology, AI, is just a core component to help make that all happen.

Tom Raftery: 33:02

Fantastic, fantastic, great. Maurice, if people would like to know more about yourself or any of the topics we discussed in the podcast today, where would you have me direct them?

Maurice Liddell: 33:12

So we have a plethora of information regarding the use of this technology on both our BDO manufacturing as well as our digital websites. So, I'll send you a link to these, Tom, so you can share this with with your audience. Great. And there's also contact information there if people wish to use that.

Tom Raftery: 33:32

Perfect. Superb. Maurice, that's been fantastic. Thanks a million for coming on the podcast today.

Maurice Liddell: 33:37

Oh, Tom, thank you. You know, it's definitely a topic I'm very excited about. I love to see how, where this technology is going. And I think you said it well, this is going to be totally different than 18 months, two years, five years. It's exciting. So thank you for having me on.

Tom Raftery: 33:53

Okay, thank you all for tuning in to this episode of the Digital Supply Chain Podcast with me, Tom Raftery. Each week, over 3, 000 supply chain professionals listen to this show. If you or your organization want to connect with this dedicated audience, consider becoming a sponsor. You can opt for exclusive episode branding where you choose our guests or a personalized 30 second mid roll ad. It's a unique opportunity to reach industry experts and influencers. For more details, hit me up on Twitter or LinkedIn or drop me an email to tomraftery at outlook. com Together, let's shape the future of the digital supply chain. Thanks. Catch you all next time.