What generative AI means for Industrial Machinery – Transcript
Chris Pennington: Hi everyone, welcome back to the Siemens Digital Industries Software Smart Manufacturing podcast series. I’m your host, Chris Pennington, Global Industry Marketing leader for Industrial Machinery at Siemens. In this episode, we’ll cover the hot topic of generative AI in the industrial machinery and manufacturing environments.
I’m joined by our industry expert Rahul Garg, Vice President for Industrial Machinery at Siemens. Rahul’s 25-year career includes delivering global software-based solutions for product engineering and manufacturing innovation. In addition, we’ll be speaking with our subject matter expert, Noam Ribon, technical product manager at Siemens software. Many thanks for joining us today, Noam. Before we begin, could you tell us a little bit about your background and give us a brief introduction to your role here at Siemens?
Noam Ribon: Well, thank you, Chris. It’s good to be here. Good to be with you, Rahul. I’m a technical product manager and I’m with the business unit that creates our digital manufacturing software solutions. And when we talk about digital manufacturing software solutions, we talk about production planning, production, simulation, production, scheduling, even production logistics. We also talk about computer aided machining and really the emerging field of what we call data driven manufacturing.
I’ve been working mostly with customers listening to their business problems, trying to see if there’s something that we can provide to assist, and if not, see if we can develop something to address that. And specifically, I have been looking into AI for the past several years, but obviously in the last year or so it’s been mostly exclusively Generative AI.
Chris Pennington: Thanks for the introduction, Noam. I have to say I think you’re working in an interesting area. Now, while generative AI is still quite new, I was reading a recent testicle by MIT that found that 64% of manufacturing companies are researching the use of AI and their own 35% of started to put AI use cases into production.
So, it seems that just as companies start to be coming to terms with AI, we have this new kid on the block with generative and what I’d like to do is hear both your thoughts on the trending topic of Gen AI and more specifically, what is Gen AI and why is it becoming relevant to manufacturing? Rahul, maybe you could start and share your views first.
Rahul Garg: Certainly. Let’s ask our Copilot what Gen AI is, and here’s a typical response:
Pre-recorded Gen AI response. So, if you were to summarize to me what’s interesting from Gen AI is the ability to create new content, I think that’s where things have really shifted and the ability to bring in new capabilities that were perhaps not explored before. And this ability of creating new content be it text, be it images, be it audio, be it software code, all of that is creating new possibilities, and to me, the biggest possibilities are around productivity improvements as well. So, I think that’s what generally is really going to be standing out very, very extensively.
Noam Ribon: Yeah, and Rahul, that that’s exactly it. I would like to add that when we talk about industrial AI and we refer to text, it also means code generation. As we all know, code or programming is used by text and I think we’ll talk a little bit about that later and correctly as you mentioned, I think what generative AI brings to the table for manufacturers is the opportunity for, I’ll use the word significant, significantly increasing productivity. And I’ll give an example for those of you on the call.
I’m sure everybody is familiar with it, but let me just brief example, not necessarily for manufacturing. Think about a document. It can be 5 pages, then pages 20 pages and you have to read it and you have to summarize it. That would take you five 10-15 minutes. With generative AI, you take this document, you post it, or upload it to a copilot. We sometimes use the term copilot. Some of you have the latest Windows installment. You have the copilot with it. You upload the document to the copilot and with the compiler will do for you. Will summarize the document. So now instead of spending your 5, 10 and 15 minutes reading the document, understanding what’s going on, the copier works in the background, and you have a summary within a few minutes. Then think about the productivity gain.
If you have to do many of those a day, so that’s just an example about the increased productivity. Similarly, we’ll see the same thing happening with industrial applications.
What really makes a difference with generative AI then? Maybe what we call sometimes traditional AI machine learning is accessibility. Those of you worked with, let’s say chat GPT or copilot, all that’s needed for you is put in a natural language prompt. That means almost or everybody can use it, and that makes a big difference. And on the other end of it, generative AI can be easily adapted to specific needs of manufacturing and other applications. And this is, I think what’s really make the difference between the AI we saw before and the advent of generative AI.
Chris Pennington: So, for manufacturing in particular, what is driving this change, the shift from AI to generative AI? Rahul, perhaps you can go first.
Rahul Garg: Chris to me I think there are a couple of factors that have come in.
Now, compared to AI’s generative AI as Noam mentioned is the accessibility Gen AI is making AI available to pretty much everyone. Previously, when we’re leveraging AI capabilities and AI technology, it was available in the hands of a of a few experts. Now with the with the advent of generative AI and the accessibility that you are that you have with it, it’s been democratized and now you have access to generative AI capabilities to the mass audience and that’s to me is the one big difference.
That accessibility makes it easy for anyone to start now using these capabilities on the shop floor, where an operator or a technician is using some equipment or be in the design engineering house where whereas an engineer is designing the equipment, all these people can now take advantage of what genre has to offer and incorporate it in that early phase. I think that’s the big difference in my point of view.
Chris Pennington: Thanks Rahul. And Noam, maybe I could ask you how mature is Siemens in its AI journey? Is this a new topic in shop floor simulation and operations?
Noam Ribon: No, Siemens is very mature. I think Siemens AI has been part of our technology journey for the last half century. I wasn’t there but really for a long time and there’s there are many AI applications integrated both in our products and our processes. The one we use internally in the company. So, I’d like to give two examples. So, 2 quick examples maybe on one end I’ll talk about the Siemens Edge device. An Edge device is a device that’s used on the production floor. Think about it as a computer, a rugged computer. It’s used to maybe collect data and analyze data, something you might sign other times do on the cloud.
But if you need to do it actually on the production floor connected to your production network, you would use the seamless edge device. The device has a built in AI model. That’s one example. On the other hand, think about production optimization. Simulation is something we provide to our customers when I talk about production optimization, the example I give is if you currently produce maybe 60 widgets an hour or 60 parts an hour and you need to make 65, then you ask yourself how do I optimize my current production line to produce those 65 parts? We have a production simulation.
To do that, you run different scenarios and the number of scenarios can be very large. You can change many parameters can be hundreds or sometimes thousands of scenarios within that program. We use another branch of AI. We call it a genetic algorithm. It’s based on evolutionary theory. The fittest, continue and those do not fit drop. Anyway, with that genetic algorithm. Again, in AI solution, we sift very quickly from those hundreds of thousands of scenarios to the maybe three or four that really would be used to optimize your production. So again, AI has been embedded in our in our products.
So, Siemens is continuously investing in AI projects. We had many in our pipeline before our call it before Chat GPT. But with the introduction of Gen AI, the number of AI projects has significantly increased.
Chris Pennington: It’s great to hear, Noam. I want to go back to you, Rahul. We’re already beginning to see generative AI write code in the IT industry and draft documentation in financial services, and of course everybody’s favorite, chatbots. How do you see manufacturers and machine builders leveraging the benefits from tools like Chat GPT and the Industrial Copilot?
Rahul Garg: Yeah, Chris, the manufacturers who are running a shop floor application or running a production floor, they are beginning to now see Chat GPT for example being used in problem decoding and trying to find out issues and then find out the root cause of those issues. They’re using it in getting good documentation, and good information on how to make some parts, making sure it’s relevant and usable, and quite a few other applications where troubleshooting of machines both the equipment and by the operators for them to communicate with different systems that can be easily accessed using Chat GPT and other such applications.
In addition, I’m also seeing now these applications becoming useful for the companies that actually make the equipment, the companies that make the machines that go on the shop floor when machine builders are now leveraging gen AI to boost their own productivity in the design on the front end design of their machines, and even in the way they actually assemble those machines, and even the way they service those machines.
So, I think we are going to see as companies as you highlighted right in the beginning, companies are willing to adopt gen AI a very aggressive manner. And I do strongly believe that this is going to be a capability that’s going to become part of everyone’s daily process and whatever task they’re engaged in.
Noam Ribon: Yeah, exactly, Rahul, if we talk about the details, AI and gen AI, maybe even together is used to troubleshoot equipment failure on the production floor so you can come back to production as quickly as possible. I’ll give the example so you can type in a question to you, your copilot or your chat bot and ask how this equipment failed. Can you collect data for the last hour or two? The chatbot, the AI will collect the data and present it to you in a way that made you make sense to try to understand. Maybe where the issue is and once the issue is discovered that semi you may be using gen AI. As you correctly mentioned, can go back to maintenance documentation, but that same piece of equipment and another point of yours will show me how to correct it or how to fix it. It will pull up the correct information from either the maintenance document or maybe for the maintenance document and some procedures that you have in the company.
So again, putting the AI together on the shop floor will allow maybe the technician that comes in to fix the equipment really get it back up and running faster. On the productivity side, maybe for machine builders again, just the details. So, machine builders many times make their own metal parts. They cut parts. That’s how we refer to it, and they cut parts using computer numerical control machines or CNCs. And typically those parts nowadays all of the parts are designed on computer aided design or CAD software and the task of the operator let’s say is to take a look at the part on the CAD system, understand what processes are needed to cut it, understand which tools are going to be used on the CNC machine to carry it, and then at the end generate a program with sometimes called a G code or M code actually to run on the CNC. Where generative AI can help is can really help our machine builders to boost their own productivity through faster ways to get from that CAD part to a complete and running program on your CNC machine.
Rahul Garg: You know that’s a good example. I do agree. I think leveraging gen AI, I for part cutting and CNC machining can be a very, very cool way to further improve their overall process and productivity.
Chris Pennington: Yeah, I agree. Really interesting. And I mean, I’m just wondering now as a key partner in product development and automation for thousands of companies, what specifically is Siemens doing with AI to create value for customers today?
Rahul Garg: So first, yeah, I’ll pick it up for an equipment manufacturer and for machine builders. Siemens, as Noam indicated, has been investing in and heavily in quite a big way. Even when you first look at one of our applications called heads, it’s used to design the machine and kind of help you optimize the best design given certain parameters that you may want to use to while you’re designing a machine. It’s using AI within the product itself, within the equipment itself.
Trying to include AI within your applications within the equipment that you’re using and using it to further, perhaps improve the overall usage and the performance of the machine. So, for example, a typical machine may have an HMI (human machine interface). AI can now be used to further simplify the human machine interface, and we are helping companies do some of those things. There’s a copilot to automate the code generation for rapidly generating, optimizing, and debugging the automation code, and enabling companies to help localize their shop floor instructions.
So, if you’re sending a machine, if you’re selling a machine to different parts of the world, you need the ability for the operators are going the machine to go to use the machine, be able to understand that. So being able to support multiple languages it looks like a very simple task, but there’s a lot of work involved in the back end to make sure that an operator in Brazil can use that same machine effectively as an operator in the US or in China. So being able to support those multiple languages becomes a very, very important answer as well.
These are all examples of how we are now supporting the machine builders and in their in their abilities to use generative AI and improve the overall performance of their and productivity of their own personal but also improve the usage and productivity of the equipment that they are eventually making available to their customers.
Noam Ribon: Rahul, these are great examples. And as I mentioned before, the number of projects exploded and we look at AI or generative AI, and now we even break it to maybe 3 categories. It’s not clear cut, but we talk about analyze, optimize and generate. I’m going to give a few examples for each one of them. And again, it’s not clear cut. You can put things in in either category, but for example analyze and we talk about productivity. It can be as simple as our application analyzes the way that you use it daily or routinely can understand what you plan to do and maybe predicts the next command. We call it command prediction.
You can extend it to even complete several steps together. Again, hence your productivity using the application. If let’s say somebody’s just ramping up or an application or checking out some new feature, why not use a chatbot just to have a how to and, ask “Hey, how do I do this in the chatbot digs into the product documentation” and provides the response for you. Where you talked about chat. What’s for production data and equipment documentation? So that would be in the analyze. There are many more examples if we talk about optimize that’s been discussed a lot, but let’s look at utilizing machine vibration data to predict failures allowing you to optimize maintenance schedules. That would be the shop floor.
We talk about manufacturing execution systems or MES systems and MES systems need to be often configured to fit the next batch of production or the next set of product needs to be produced. Gen AI can look at the information on the MES and look at the next production monitor. What’s going on? Gen AI offers the next configuration so that can be done easier and much quicker, and we talk about energy utilization to improve carbon footprint. Take your new production plans. Take your past energy, spend data and use AI to optimize your throughput to optimize the energy usage to save carbon footprint. Again, with Gen AI. And as I mentioned, I mentioned now throughput. So, as we talked about, you can optimize throughput of production as before and the last one, which I think will generate the most productivities is generate. Rahul, you mentioned already industrial Copilot for program PLCs. The same can happen for helping programming robot programs.
We can use a Copilot for scheduling production. For example, If I need to set how parts or how material goes through machines, that’s scheduling production, maybe or generative AI can test a few scenarios. A few options and proposed the most optimized way to schedule production.
Chris Pennington: And I’m sure gen AI’s going to help to drive a lot of standardization and things like PLC code, because of course you know you’re getting standards built into the coding. So that’s a really exciting will way forward and it’s great to hear all these ideas of what actually is happening today. But what about if we look in the future a little bit? I mean, maybe you could give a picture of what the future might look like for Siemens and other manufacturing enterprises with the assistance and widespread use of gen AI.
Rahul Garg: Actually, this whole analyze, optimize and generate is actually quite a good way to construct the overall adoption of generative AI. To me, where this further really brings it home for us as Siemens is with our teams that our application. So, when you think of generative AI, one of the most important capabilities that you need to adopt gen AI, this is a lot of data and that’s where the whole concept of large language models comes in as well. And Teamcenter is probably one of the largest world’s largest repositories of product data and the advantages it’s highly structured. It’s hierarchical in nature and it can traverse that information very easily.
So, leveraging gen AI with Teamcenter becomes a very valuable asset for any company that’s trying to figure out how to unlock their product information in a more effective way and making it more easily accessible not only within the company or and the users of teams and or like the engineers and the designers, but also to the extended audience of all people who are interacting inside the company with the equipment beat on the manufacturing shop floor, be it in the service department or even the end users of the machines. So having teams’ scenarios as an important resource as a data resource makes it very valuable for customers to now further leverage the value of generative AI.
Chris Pennington: Some great points there, Raul. To from what I’m hearing, there are some exciting times ahead for manufacturing. Noam, could you give a picture of what the future might hold for Siemens and other manufacturing enterprises with the assistance and widespread use of gen AI?
Noam Ribon: Yeah, the future. Lots of enthusiasm as you’ve just heard. For me specifically as a technical product manager is about the near future when our customers are starting already using our generative AI applications and seeing the productivity gains or the benefits. I can tell you that I shared a few projects with several customers and then the project was very well received, lots of anticipation of being able to use it as soon as possible.
Rahul Garg: Yeah, I would say, Chris, from a future perspective, generative AI has the potential to even revolutionize the way companies design, develop, manufacture and operate equipment rights and it definitely has the ability to have a big impact the way we are using the whole concept of a copilot to help assist in that process. And make sure that we don’t take our hands off the wheel, if you will. And we’re using the copilot as a support function to help us in in, in our in our abilities to design, develop, manufacture, operate the equipment.
But as the usage of generative AI improves, and as we bring in more and more capabilities, I would see a bigger and bigger adoption to me. Where I see the big a shift also compared to using gen AI in in what I would call as that nonindustrial world is many a times a good enough answer is also good. Whereas in our case, when you’re manufacturing something, a good enough answer is not good. You need it to be precise and that’s why the copilot process makes it in the short term and in the near term more valuable. And as uh, as the adoption of gen AI improves, then the potential to, as I said earlier, to revolutionize the way things are done.
Chris Pennington: Thanks for that, Rahul. Now it’s been great to hear both your thoughts on this topic today and I appreciate the opportunity I get now to share your insights and vision with our listeners. But as usual, before we end the conversation today, are there any final thoughts, ideas, or advice either of you would like to leave our listeners with?
Rahul Garg: The one thing I would say is this is a fast-moving train, and you don’t want to be left behind waiting for something to be perfected and then try and use it. I would urge everyone to start that journey. There are many applications that are already available today that are in use and in production with many companies. Try and explore that. Get it going inside your company. Get some pilot started, get some application started and the faster you get going with this the better you will serve your company as well.
Noam Ribon: Yeah, Rahul, all I can do is simply second what you said. I would as well encourage you to not delay examination how AI can boost your organization’s productivity today. Set your goals. See where you wanted to take you plan, how to measure results, but really start experimenting.
Chris Pennington: Thanks, Noam. That was a great way to finish today’s podcast. It’s been a pleasure having you both as our guest today and especially Noam Ribon our special guest for coming here. Thank you both for joining me to discuss the significance and generative AI in manufacturing. And I’d also like to thank our listeners for tuning in today. We hope you stay tuned for future episodes where we’ll continue to explore the future of manufacturing. Thank you.
Siemens Digital Industries Software helps organizations of all sizes digitally transform using software, hardware and services from the Siemens Xcelerator business platform. Siemens’ software and the comprehensive digital twin enable companies to optimize their design, engineering and manufacturing processes to turn today’s ideas into the sustainable products of the future. From chips to entire systems, from product to process, across all industries. Siemens Digital Industries Software – Accelerating transformation.
