A Forward Look on AI in A&D Part Two – Transcript
In this episode of Talking Aerospace Today, Todd Tuthill and Barclay Brown, Associate Director for AI research for Collins Aerospace and leader for the AI Systems Working Group at INCOSE, continue their conversation on how AI could change the future of aerospace and defense.
Patty Russo: Greetings, everyone, and welcome to another episode of Talking Aerospace Today from Siemens Digital Industries Software. I’m Patty Russo, and I’m responsible for global marketing for our aerospace and defense vertical here at Siemens. Thank you for joining us today. In our previous episode, you might recall, Todd and I had the privilege of speaking with Barclay Brown. Baclay is Associate Director for AI Research for Collins Aerospace. He is also the leader for the AI systems working group at INCOSE. Barclay joined us to talk about the future of AI in aerospace and defense. Barclay is an expert in his field, evidenced by the book he wrote on the subject entitled, Engineering Intelligent Systems: Systems Engineering and Design with Artificial Intelligence, Visual Modeling, and Systems Thinking.
Patty Russo: In our last episode we talked about how AI could change the A&D industry in the short and long terms. We addressed the topic of trust related to AI and the full extent of where this technology could be applied in engineering processes. So let’s just hop right back into the conversation with a question for Todd.
Patty Russo: You covered things like the concept of AI and the need to move from, “Hey, this is really cool,” to trusting it and talked a little bit about the functionality and the hardware and then of course the example you just gave Barclay on digital engineering. The question that that tees up is, and Todd, I’m going to ask you this, what types of aerospace related innovations or new products or opportunities can this type of technology, can this AI technology in the practical application space, maybe digital engineering or other areas of our industry, where can this play and enable those innovations in the future?
Todd Tuthill: Yeah, I think there are several. The first few that come to mind are some problems we’ve been trying to solve for a while, but just haven’t got there. We want to travel to Mars. We want to be able to live on Mars. There’s technology we’ve developed to be able to get there. There’s some that we haven’t. And as Barclay said, and as I’ve said as well, one of the things that AI is going to do, it’s not going to invent something necessarily that’s brand new. It still takes human engineers and their creativity to do that, but it’s going to help us get to those innovations like traveling to Mars or maybe like some new propulsion system to solve greenhouse gas. It’s going to help us get to those things faster, so that’s the first thing I think about. Another thing I think about, again, back to the holodeck example is imagine a way to collaborate. You know, in Star Trek, they’re all physically there together.
Todd Tuthill: The use case I like to think about is I’m in different sides of the globe, and I’ve got this global company I work for, and I want to enter a room where I can see you and you can see me. You know, Barclay and I are on teams right now. I can see a 2D Barclay, but wouldn’t it be great if Barclay and I were in the same room? We could walk out to the flight line, see the jet there. You know, the virtual flight line. See the jet we’ve just developed and interact together as if we’re in the room together and not have to travel all the time. And so I think it’s going to enable incredible new ways of collaboration that right now we can hardly even dream of in terms of the again the visuals, and then when you talk about you know connecting those visuals because there are technologies today.
Todd Tuthill: Like Barclay was talking about to create video, really high resolution video, but there are very few connections right now in the digital engineering world that let us connect all the physics behind that. You know the example of Barclay talked about. Gee, I want a bigger aircraft. I want a half meter bigger wing. I could do that today with some more technology. What I can’t do today is instantly do all the engineering and design. It would take to actually put all of the hardware and all the analysis and all the aerodynamics behind all that, in an instant. Those kinds of things, we have technology to do a lot of those things, but it’s hours now, not seconds. And this technology and the software technology and the hardware technology I think is going to get us from the hours and days to do that optimization to seconds. So that’s coming. And then once we’re there, then we can start dreaming about all kinds of cool new things.
Todd Tuthill: I think about autonomy, and the autonomy that we could come up with if we could get there in terms of autonomy for transportation. We talk about advanced air mobility and lots of the problems we’re having there. I think it can help us solve some of the hardware problems, but I think some of the bigger issues are going to be within air traffic control, within the safety aspects of autonomy. When can we get to a place where I can really, really trust autonomous cars, autonomous aircraft, autonomous spacecraft to fly me all kinds of places around the Earth, or maybe to other places with these new propulsion systems we’re going to create. And that’s the civil side. I think we’re seeing that a lot in defense. The age of the manned fighter I think is starting to come to an end.
Todd Tuthill: When you look at NGAD and what’s happening with NGAD and the collaborative aircraft, that’s where you’re going to start seeing a lot of drones flying with manned aircraft. AI technology embedded inside those aircraft are going to become. things will be going to become more and more autonomous and do more and more things. And then we can talk about manufacturing and all the really cool opportunities that there’s a bunch of really dangerous things that human beings have to do right now in, in manufacturing big products that AI is just going to give us an opportunity where real people don’t have to put themselves in harm’s way to manufacture products anymore. And the list just goes on and on and on. And I mean, what would you add to that Barclay?
Barclay Brown: Yeah, it’s fascinating. I mean, it’s, there’s all kinds of room for creativity. I mean, I think one of the things you know is we’re thinking about, like, well, what’s going to, you know, is this just going to happen easily? Smoothly? Are there any barriers or any risks, you know, to implementing this, right? And I would say one of the things is we’ve got to, as engineers, become more conversant withtechnology that’s driven by software. And that sounds funny because you think, well, engineers will be all hip to the latest software stuff. But as you may know, there’s a lot of engineers that aren’t actually so hip to software. You know, they’re like, “What are you talking about, software? I’m an electrical engineer. I’m a mechanical engineer. I’m a civil engineer.” Right? Somebody the other day told me, “I’m going into civil engineering, so I don’t have to worry about any of thisdigital or AI stuff.” You know? And I’m not sure that’s exactly the case, but that was this person’s perspective.
Barclay Brown: Right, so I recommend that people learn Python, just to put a fine point on it, right? So engineers of all disciplines have to have the ability to use software in their own work to connect things, to do things. They can’t be any more like, “Oh, that’s going to require some programming. So I’ll call up IT and hope that they do my project so I can continue my work.” You know, it was like in the day when spreadsheets first came out, people were probably saying, “Well before to do to get any financial calculations done, I just call up the accounting department and they do it for me.” And what a breakthrough it was to now all have Lotus 1-2-3 or Excel, right, to do calculations of my own right here on my own desktop. And so we’ve got to get to that with digital engineering and with software.
Patty Russo: Bringing it back to one of the things Barclay that you had touched on a couple of minutes ago, which is something that you’re going to be talking about, the digital engineering. Todd talked a little bit about what he thinks AI can advance in the industry. I’m curious, Barclay, from your point of view, obviously you’ve talked a lot more in depth about how it works, but what do you think, Barclay, that our industry, our audience, maybe could take away from this in terms of what they can think about doing with this technology in a practical application? Perhaps in the short term, obviously long term, we can pontificate and these are all great, you know, great things to think about. But you know, at the end of the day, our audience probably has a job to do and they’re thinking, “Okay, how can I make this work for me either today or in the short term?” What are your thoughts?
Barclay Brown: Yeah, it’s a, it’s a great question. The, you know, the place to start would be and, you know, there’s a lot of engineers that are already thinking this way. I mean if you if you show an engineer something like that, GPT, they’re immediately going to be thinking about, “How could I apply that kind of technology in my work?” Right? And so what are they going to come up with when they think about these things? Well, most engineers, and I think systems engineers are even more of a case this than in this than some other kinds of engineers, if they split up their job and I say, “Well, here’s the different things that I do every day.” Some of those things will be things that they know they have to do. It requires their own creativity, their own thought, their own knowledge, their own background, their own judgment. That’s a big one.
Barclay Brown: Large language models have no judgment, right. But then there’s another part of their job, and maybe it’s even a large part, where they’re not having to think very much. They’re rearranging information into different forms. They’re summarizing this so they can put it in this form and put it over here like this and add it, correlate it with this so they can show it to someone else and changing the format of this and recalculating this to reduce it to that and reformat it so they can feed it to this other piece of software. All that kind of stuff is ripe for automation, and automation using AI, automation just using regular old software, and all of that. So that’s one major dynamic is people have to start thinking that way. Now for us to really think that way, we need another major change.
Barclay Brown: So in addition to engineers getting conversant with software and with software-oriented ways of doing things, there’s another big piece and that’s data. We think about data in our corporations wrong. Maybe wrong is too strong a word. I don’t know, we’ll see. But we think about it in a way like, “All right, the data is closed. It’s my little set of data. I own this data because it’s my data because I use it.” Well, for one thing, it’s not your data. It’s the company’s data, right? And we our default is to not share data. And sometimes this is driven by a need-to-know kind of sentiment, right, where I shouldn’t share things unless there’s a super compelling necessary reason to share it. But somehow that’s got to get tempered with the need to share things so that the enterprise can treat the data as its own asset and start to use it and not have data locked away in corners.
Barclay Brown: Have data being stored more in shared resources that I can get to if it’s being indexed and locatable and where I can use it. And this is one thing large language models have opened up tremendously, right, is because they have opened up access to text data, that I now can store free text data and have LMs make sense of it, find it, retrieve it, reformulate it to all that which that wasn’t possible before just a few years ago.
Patty Russo: And that kind of lines up, Todd with what you were talking about in terms of collaboration, where you see this going. Do you want to expound on that at all based on what Barclay just said just in terms of how will unleashing, I’m going to call it unleashing the data, how will that help facilitate more efficient, more effective collaboration?
Todd Tuthill: Yeah. And I think we can talk about that at several levels. Barclay talked about a great example earlier about spreadsheets and how spreadsheets kind of made accountants obsolete. Well, regular listeners of the podcast here will know how much I disdain spreadsheets and how much I want to see digital transformation. get rid of spreadsheets. So I really want to want to think about it and use that analogy. When do we get rid of the spreadsheets? Because there’s data in spreadsheets. It’s locked away. So the first level is let’s find a way to get all of that data in a place where we can all access it.
Todd Tuthill: And I think we have tools today to do that, but the problem is the vast amount of that data make things really hard to find. And one of the really cool things about AI and large language models, other aspects of AI is that they can go process all that data very quickly and they can make things very easy for us to find an access. So that’s the beginning level of just finding the data, but then, and we haven’t talked about the industrial metaverse, but that that’s kind of implied in the whole holodeck thing. But then if you start taking that data and mixing it with, you know, augmented reality or industrial metaverse or whatever you want to call it, you start to have better ways to interact with that, and let’s stop thinking about interacting with that data via keyboard and a mouse and start thinking about a new way of the future of interacting with our data is large language models.
Todd Tuthill: I’m not going to type out a command anymore. I’m just going to ask it. Make the wings half a meter wider. Right? That’s a new way of interacting. And that’s the next level of how we can interact with and use our data. But those things are all still local, and then the whole thing about the holodeck and the virtual holodeck around the world, that starts to get us to places where now I can interact with people in better ways. And you can just go on and on and build it. And it’s again something that Barclay said, I think early on is key. He said this won’t happen all at once. It’s going to happen in pieces. And those are some of the pieces that lots of companies are working on. Some of those pieces are available today.
Todd Tuthill: But that I’d encourage listeners to think about. What are some of the steps I can start taking about my data, interacting with my data in better ways, retrieving things better, having higher level ways of thinking of retrieving it and interacting with each other? And it’s a step at a time and we all take those steps and we’re going to look back in a few years and say, wow, I don’t use the keyboard and the mouse anymore. I talk to my computer like Scotty did in Star Trek. It thinks, it does all the busy work, the mundane stuff for me.
Barclay Brown: It’s a great point, and in fact I was going to go back to, since you brought up spreadsheets, right, so I’ll do the kind of the counterpoint to this. Maybe, right. So I invented this little course a couple of years ago, and it’s called Excel with Python for engineers. Okay. And it’s this funny little course. It’s a four hour course. And the idea was I can teach people who have no programming background just enough Python, just a little bit, not the whole language, just a little bit of Python so they can do one particular type of application as follows. Read data in from a spreadsheet, do some things with it, add other data, compute, add, blah blah blah, manipulate it, and then spit out a different set of data to another spreadsheet.
Barclay Brown: Spreadsheet-based, right? So this is not the ultimate solution, but this is a today thing, right, that I can actually automate stuff even though the data leaves lives in spreadsheets today. Maybe it’ll live in a better place later in a database or in a model or something. But today I can include that stuff because I don’t care where the data is. I just need to have access to it so I can do stuff with it so I can do what I need to with it.
Patty Russo: Yeah, the Excel with Python for Engineers 2.0 will probably be the next iteration, but probably because of collaboration and input from the users that took your course because they’ll probably find a way how to make it better. So with that, I’m going to ask kind of an open-ended question here as we reach our time. Is there anything else that either of you would like to comment on in terms of AI where you see the industry going or perhaps practical applications that are possible today before we wrap things up?
Barclay Brown: Well, I’ll go. I, you know, what I would tell someone is, you know, embrace all of this, which sounds a little simplistic, but really no kidding. There’s nothing to be afraid of here, and an engineer at any age, at any stage can learn enough to be able to incorporate this kind of stuff. And it truly is fun. What’s really sometimes really uncanny about large language models and applications using large language models is they don’t rely on programming knowledge. We have teams creating large language model applications and only some of the team is writing code. But the other parts of the team are figuring out how to talk to the large language model, how to prompt it, how to get answers from it, how to incorporate it, so on. So it’s broadening the set of people who are going to be collaborating to create what we might call an application or a usage and stuff. So it’s going to involve everybody.
Barclay Brown: And in this stuff, another thing is great about being around today in this kind of a change is the learning resources are endless. You know, online courses, some of the best education you can get is online and for free, free or cheap. Right? And you know, there’s a bunch of free courses on deeplearning.ai. One of my favorite sites, Andrew Ning, one of his creations. They’re free courses on large language models and you can learn more than most people even imagine about large language models right there in these little courses taught by the industry leaders for free. You know, how do you beat that? Right? So the knowledge is there, it’s available and it’s fun to get into for any engineer.
Todd Tuthill: Great things, Barclay. I want to end with another Star Trek story if I could, and I’ll date myself. I don’t remember the name of the movie particularly, but it was released in the 80s and basically the crew have to go back in time to what was then modern day San Francisco to rescue a whale.
Barclay Brown: The Voyage Home.
Todd Tuthill: Yeah. Barclay knows Star Trek better than I do. And they got to go rescue a whale and they need to take a whale in the Enterprise back and forward in time. And I remember specifically, sitting in the theater, Scotty picks up a mouse on a computer and starts to talk. He says “Computer, do this.” And he’s mystified as to why he can’t talk to his computer. And again, this is the 80s now. I remember sitting in that theater and thinking, will there ever be a day in my lifetime when I can talk to my computer with natural language? And I think about that, I look at my iPhone, I look at the iHome devices in, in my house, I talk to my computer all the time every day. That day has arrived and the technology that a few, just a few decades ago I thought was absolutely impossible. I use every day as if it’s nothing. AI is all around us. We’re using it every day.
Todd Tuthill: It’s going to come faster and faster and faster. And it’s going to make our lives better because I think my life is a lot better now that I rarely have to ever type text messages on my iPhone. There’s all kinds of great applications of AI today. There will be many, many more in the future, and I just encourage people to learn more about them, use the resources, Barclay said. And to get involved because I really believe. That this is going to make our personal lives and our working lives much, much better in the future.
Patty Russo: This adds another level about AI and the industry and is a great place to wrap up the discussion. So big thanks to our guests, Todd Tuthill and Barclay Brown for their insights today, and thank you as always to our listeners for tuning in. If you just joined this conversation, I do invite you to listen to previous episodes of our talking Aerospace Today podcast for more inspiring conversations like this one. We look forward to having more discussions about exciting opportunities and innovations ahead, as well as the advances that are being made in the aerospace and defense industry. I’m Patty Russo and thanks again for joining us on Talking Aerospace Today.
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.
