Digital Transformation industry leader Jeff Winter joins 4IR Solutions CEO James Burnand and CTO Joseph Dolivo on “Heads in the Cloud” podcast to discuss Pt. 1 of Industry 4.0 and AI trends.

‍

James Burnand:

Hello everyone and welcome to another episode of Heads in the Clouds. I'm James Burnand. I have with me as always Joe Dolivo. And today we have an extremely special guest, Mr. Jeff Winter. For those of you who don't know, and I don't imagine there's many of you out there who haven't seen or heard of Jeff before. He's the number one industry thought leader according to on Analytica, as well as many other prestigious awards. He's the chair of committees, he's a part of a lot of different online statistics and information that's provided. And bluntly, he's probably the most prominent leader in Industry 4.0 that I know. So I wanted to say hello and welcome to the podcast, Jeff.

Jeff Winter:

Well, thanks for having me. I am excited to be here, especially because of my unique relation with both of you guys.

James Burnand:

So full disclosure is both Joe and I have worked with Jeff in the past and we have a fairly long-standing professional relationship, as well as I consider Jeff to be one of my personal friends. So I'm super happy to have convinced him to come on our little podcast and hopefully for anyone that's listening, you can get some value out of this conversation.

Jeff Winter:

I'm looking forward to it.

James Burnand:

Cool. So we've seated up a couple questions just to try to make sure we can drive the conversation the right way, so I'll get started. So Jeff, what do you think are the biggest emerging trends in the industrial space this year?

Jeff Winter:

So this is a funny question, because I would have an entirely separate answer to give you if ChatGPT didn't get released, but because ChatGPT-

James Burnand:

What's ChatGPT? I've never heard of it.

Jeff Winter:

It completely dwarfs everything. Never heard of it? You should ChatGPT what ChatGPT is. So this technology, and keep in mind when I say the word ChatGPT, I'm generally referring to all of generative AI, but ChatGPT has become the primary word like Google to describe all internet searches. So when I say ChatGPT, I don't necessarily mean OpenAI's ChatGPT experience that you have, but all generative AI, because generative AI has been out for years, but it really became publicly known as a household term basically at the end of 2022.

And since then it is taken off really by storm. It's impacting every industry and I would argue every function from the entry level all the way up to CEO. And what makes this technology so different than other massively disruptive technologies out there is an example like blockchain.

I would argue blockchain is a massively disruptive technology, but blockchain takes a lot of time, energy, and money to be able to fully utilize and implement. ChatGPT is one of the fastest technologies that you can implement. And since Microsoft made their investment to OpenAI and they get access to the models they've publicly announced, they're integrating every single one of their products. So now it's just going to be a feature that's turned on where you don't have to do anything as a user. And that's what makes this technology so much more disruptive of everything else, is how easily everyone can use it. You can implement it in hours instead of years.

James Burnand:

I know, I watched one of the recent announcements and talking about Microsoft co-pilot, where Satya and a bunch of the different folks from Microsoft were describing the future of work and how impactful it could be on the monotony of certain tasks in certain things. And actually seeing it in action applied to things like writing proposals and documents and gathering... One of the coolest things I thought was how it could summarize meetings and take the key takeaways. And there there's really a ton of, I'll call them office centric practical use cases, but what do you think about manufacturing? How do we apply it to the actual manufacturing process? What's the biggest use case there?

Jeff Winter:

I think everyone is trying to figure that out right now. The other part that makes this technology so disruptive is the fact that everyone is trying to figure out how best to utilize it. Think of it as just a creativity enabler. And those that are able to harness the power of this creativity and apply it to different aspects of their business faster will be in such a huge competitive advantage over others.

So a lot of what I like to talk about and used to talk about with various different clients is around understanding how to think about it, so that you can most appropriately apply it to your very specific business cases, because there's so many opportunities out there that you don't have to look and go, what's the number one use case that another company use. You can look at as to what would be impactful for your business based off of what data is readily available.

So one of the things you have to note is that really there's two main ways that you can think about ChatGPT as a concept. One is the fact that there's the public version using its predetermined large language model to understand what you're typing in and to spit out an answer. And you can think of it as having its huge corpus of data based off of what the internet knew at the time that it was generated. That's one way to use it. All right, just how to use it to help speed up your processes based off of common understanding of just normal things. The second is linking the model to your data and having a ChatGPT for your company based off your own data. That one though, you can't really do if your data isn't properly normalized, contextualized and properly indexed.

So it ends up driving a lot of companies to go, what data is most important for our company to have access to in a ChatGPT? And we need to get our systems and even data structures set up so that we can properly use it. So to answer your question, what I'm seeing right now that most companies are looking to use quickly, is what they already have digitized and readily available. As an example, I have some companies that have their user manuals for their machines already digitized. If you have your user manuals already digitized and they're properly loaded in, you can feed them right into a ChatGPT model and ask questions like, can you summarize the maintenance procedures on this entire line of those 12 machines, and write out a safety checklist in both English and in Spanish, and produce a quiz for my employees to make sure that they understand how to use it. You can do that today if you have the data.

James Burnand:

Well, that's a really cool use case. And I honestly think a lot of these, the issues we have with ChatGPT is the creativity of knowing what to ask for. But going back to your point at the beginning of that, the last part of the discussion, talking about data and talking about clean data, organized modeled information that can be useful. I think there's such a huge demand for that now, and we've seen so much of that as being a baseline requirement for so many things, including these AI tools. I wonder if this will finally be the thing that really drives the organizations in the manufacturing space in general to be investing in this completely and heavily, because I'm seeing movement in it, but it's still not consistent. You got thoughts on that Joe?

Jeff Winter:

I think you're right. Were you saying me or Joe?

James Burnand:

Joe hasn't got to say anything. Joe hasn't got to say anything yet.

Jeff Winter:

Go ahead Joe.

Joseph Dolivo:

No, I'm enjoying this and I wanted to hear your response Jeff, and then I'll jump in with my own snarky comment about Clippy, so please go ahead.

Jeff Winter:

All right, I like it. So I would say that what you're going to find is that ChatGPT is going to be the driver for a lot of other initiatives and it will re-change the priority of some existing initiatives, good examples being the proverbial IT/OT convergence. This may completely shift its priority in the organization and what parts of it you're actually focusing on, because of the use of ChatGPT, even though ChatGPT itself is a small investment and a fairly easily easy thing to roll out. The big heart of the problem is going to be getting your data readily available. And I think this is going to completely shift what companies do with it. There's one question I usually like to ask leaders of companies, and this shows two different things simultaneously.

I usually typically ask leaders of organizations, if you had ChatGPT readily available on all of your company data, in theory, what would you ask it? And when I do that and I find the CEO or the CIO, the COO, the CTO, all these executives, when I get these answers back and I've done it several times, you find out really two things. One, you need to educate them on what ChatGPT is in generative AI in general, because they're not understanding it right to know what it can do and what it can't.

And number two is half the things they're asking don't require generative AI at all. We could have done it with machine learning 10 years ago and be able to produce the answer that they were looking for. So it's a big shift in education and the ability to kind of guide the prioritization of projects.

Joseph Dolivo:

I was going to say, I think what's incredible about it is that you're building on technologies that have existed for a long time in some capacity, but I think what ChatGPT has done is really bring it into the public vernacular and is exposing people to it who otherwise probably wouldn't have been with that. As with any new technology change, there's always this sort of gap in understanding and education that needs to be closed and there's also the potential for people to let's say misuse technology.

So that's I think a whole interesting discussion into itself. But you look at ChatGPT for example, it is very confident in what it says, is it right or is it wrong? So if you ask your company for example, you've got all this data, you feed into the model and you say, who should I lay off for my company, for example. Is it going to give you an answer that you would be able to stand behind? Is there going to be any human in the loop for making decisions around that? So it opens up a can of worms and it's great that it's out there now, because it's driving these discussions, but there's a whole bunch of other things, ethics aside, that I think we have to think about as an industry. But it's incredibly exciting, no doubt about it.

James Burnand:

Well, I think there's ethics and there's security. I forget the exact statistic I read, but there's already a amount of confidential data that's been fed in to ChatGPT that people are unfortunately unaware of, because their employees are using it already for helping them reword a proposal or an email or format a document. And to do that, they're actually sending in data that's being incorporated into the model that was never intended to be done that way. And so that's one example of security. I know Jeff, you you've probably got a whole bunch of other examples of where security needs to be a consideration as well.

Jeff Winter:

Not just security, but I would say proper use, to add to Joe's points. One of the first things I recommend customers do today if they already haven't done it is come up with a ChatGPT corporate strategy that includes governance around how and where to use it. Because even knowing things like, if you use the public OpenAI's ChatGPT, any information you put in there is accessible to the public is something that I don't know if most companies are fully aware of, because I guarantee most of their employees are playing around with ChatGPT. The other thing is once you use it inside your company, you also don't want to have it hinder productivity, because everyone thinks that this new cool shiny toy is able to help them, and then they end up playing around with it and asking unnecessary unrelated questions that could end up getting them in more into trouble than can actually help them.

And so having the education to know where to properly use it and how to put guidelines around is extremely powerful. And when you make your own company ChatGPT, you can train the model to say, I don't know the answer. That's entirely possible on your own data. But as an example, one of the things that I guide people to have is make sure you're using it the right way that you should be using it, and make sure you don't use it in the areas is not good at. Great examples of what it's good at; summarization, semantic search, which if you're not familiar with that, that's the ability to understand what you're asking and to provide or look up an answer for that, rather than a literal typing in exactly what you asked. It will go, I know what you meant, I'm going to look it up.

Third one is coded. That's a big one, and then fourth is content generation. Those are like four safe bets that you can use that really won't get you into any trouble. The ones that you can get into trouble is when you start to pinpoint very specific answers, especially that are based off of calculations. So anything that's numerical, you ideally don't want at least today, ChatGPT to be the thing that makes that answer. You'd rather use a different thing, which is most likely machine learning to create the answer of which ChatGPT is just looking it up. So it's an understanding of where you should and shouldn't be using it, because if you start using it over here, you're going to get into a lot of trouble.

Joseph Dolivo:

I love when you combine, like you said, concept generation with summarization. What's going to happen is you're going to be writing an article and somebody is going to put in a prompt and say, "Write me an article about this," and then somebody is going to ask ChatGPT to generate that. Then somebody else is going to ask ChatGPT to summarize that. And it's like you're creating content and summarizing content and you're left with maybe the same thing before and after, but there's going to be a lot of this sort of, let's say noise that's generated as a result of that and then a need to filter through the noise like we do with other kinds of signal processing. So again, interesting side effects of this great technology.

James Burnand:

I think it really does create a different way to do work too, is if you can... And I think this goes back to some of the material that's been released around, I'll call it monotonous task management and automation. There's a lot of things that knowledge workers in particular do today that are monotonous. And I would argue that a lot of things that are done by factory workers, as well as anybody inside of the workforce has a certain degree of monotony in their tasks. And when those tasks are predictable, manageable and quite simply need a review to be able to validate the results, so that they can be submitted or accelerated, I see that as a fantastic use case, but it changes the way we do work.

It changes the amount of effort that's involved in doing tasks and bluntly, without proper management of it inside of your company, it won't work very well. You'll have a mix match of who's using it and who isn't, and you as the company are going to lose your competitive edge if your competitors are using it to manage and accelerate their abilities for the things that it's really great at. So I think it's a fundamental shift in the way business is done. It's just exactly what that looks like is to me, it's still a bit of a mystery and I guess we'll have to see as the world progresses and these models continue to grow.

Jeff Winter:

And that's what I'm enjoying is hearing how all the companies I've talked to are thinking about applying it and using it in their organization. As another good example for manufacturers, this could be changing the way that you train your employees and give them access to systems that normally require a lot of training to understand how to use, whether it's MES, QMS, your ERP, it could be your supply chain planning.

Now imagine just having a little prompt that you need no training on, that you can ask a question like, who's the truck driver for this route? Who's the operator on this shift? You don't need to look up or understand how any of those systems work to be able to now access them. So really this could completely change the way anyone in the company goes to a computer to look something up or ever has to enter anything in. If you do either of those two things, this will completely change your life, which is most people I would argue.

James Burnand:

No, I completely... It is a huge deal and certainly when you look at the use cases and how we can apply it to the manufacturing space, I think it's probably... It's funny, because a year ago we were talking about the industrial metaverse, and I still think there's value in digital twin and there's value in co-location and virtual reality and all of these technologies that can be complimentary. But bluntly, I think they're way overshadowed by what ChatGPT is going to bring to the table.

Joseph Dolivo:

Agreed.

Jeff Winter:

Agreed.

James Burnand:

I still think it's cool to have a headset on and walk my plan, but hey, that's... So what do you see, Jeff? What do you see the next couple of years being... Obviously, ChatGPT and large language models are part of that future, but what else? What else is in your vision of where digital transformation is going in the next couple of years, and how are companies going to adopt and adapt?

Jeff Winter:

So one part I want to add to the ChatGPT which will go into this, is there's two other ways to look at this. I had it from the inside, but you should look at it when you build it on your model, so there's the public, I'm going to call it the use for language understanding, and then there's the build your own ChatGPT. When you do that, there's really two aspects. There's internal optimization. You're helping to approve efficiency on things you're already doing. You're just doing them better, faster today.

The other's completely changing the way that you engage with your customers, and that's a one that will drive a whole bunch of other changes in the way that companies work and what technologies they use. Imagine, for example, you now make a chatbot that is so effective for your company and it can be whether it's on a website, it can be the phone recipient that you call and talk to, that it's actually better than a person, because it can answer better and faster whatever you want to know about anything, any question that you have about that company's product or what they do or how they're set up or who you can talk to, that will completely change the way that you engage with people.

As an example, let's say you're a fashion company selling clothes, and you go to the website and go, I really like what Thomas Crown was wearing, what Pierce Brosnan was wearing in Thomas Crown Affair. Can you select an outfit to match that? And it can go of course, and it can select all that for you. That can completely change the way that companies engage with the world. So that I think is going to change a lot of other things. They are tied with it. I actually think ChatGPT is going to broaden overall AI. You're going to see language processing takeoff, you're going to see computer vision takeoff, machine learning takeoff, all these other things.

And if you even look at the Google trend search, so ChatGPT took over in terms of all digital transformation technologies, which I tracked something like 70 of them, but AI grew almost as much at the same time. So that means that people are going back to something that they've been talking about for 10 years and are looking at more ways to utilize an existing technology. Then when you go one level below AI, because AI is the level of making decisions and predictions for you is, well, how do we now get more of that data, which is driving IOT and digital twin Up, which is why we're seeing a huge increase in digital twin and IOT coming at the same time.

Well, how do we capture all that data? What's going to be actually making it? And so you're seeing a lot of these other technologies be pulled through as a part of it, because they became more important because their value got higher. I don't know if that answers the question, but it's really impacting all of them.

James Burnand:

I'll let Joe answer. What do you think, Joe?

Joseph Dolivo:

Well, I think it's well said. And when you talk about data, there's different kinds of data. So we've been talking about ChatGPT, which is an interface to text and speech. But you've also got numerical data, you've got sensor data, you've got all these other things of which there have been machine learning models for a long time that are being used, and there's some pretty cool applications of doing the canonical example being predictive maintenance.

But the problem is the same and that you're collecting data and that data needs to be stored in some a way. It needs to be stored efficiently. Needs to be cleansed. It needs to be put in a format where it can be transformed and put into tools that are going to provide you these insights. So whether it is a custom GPT chatbot or it's some other tool running in the cloud or at the edge, it's another interesting transformation, if you will, to avoid using that word again.

So I think in general, the appetite that folks in general are going to have for things like AI being more of a forefront is exciting. But I'm app approaching all this with a healthy level of skepticism, I think, to see what is the impact going to be and is that impact going to be consolidated only into the companies that are able to get all of this data. A company like Microsoft for example, that has the graph API and this data across all the different users. Is it going to be also something that smaller companies can adopt and be able to create a competitive advantage out of as well?

Or even just individuals to use in our daily lives is that, are we giving all this data to big companies that are going to be able to use it and provide us with insights, but then who really owns that? So I really like the other side conversations that it creates even outside of what are the possibilities we can do with this data, but just the conversations around privacy, maintaining that data and the different cultural changes that it's going to have. It's definitely transformative though.

James Burnand:

Well said and I agree. I think it's a big deal when South Park does an episode on it, I guess is my-

Joseph Dolivo:

Yep, you've hit the pop culture mainstream.

James Burnand:

And it is actually on my to-do list to see if it actually works to have my responses to my wife be generated by ChatGPT and to see if that improves my relationship. We'll see. So on that note, I think I would like to say, Jeff, thank you for being a guest on the podcast. We appreciated your time and love the commentary. We'd love to have you back another time if you're willing and able. And we look forward to talking with you again soon, hopefully.

Jeff Winter:

Awesome. Looking forward to it.

‍

‍

Connect with one of our experts: https://info.4ir.cloud/connect-with-our-experts  

Subscribe

Audio Podcast: https://spotifyanchor-web.app.link/e/gzLK3xcPMsb

Blog: https://www.4ir.cloud/blog

‍