AI in Design: Coming Full Circle

As a Generation X engineer, I’ve witnessed remarkable shifts in how we approach design engineering.

Recently, I saw an article suggesting Gen X is frustrated because the skills we learned early in our careers no longer apply in today’s technological landscape. This characterization made me pause and reflect. While our tools have certainly evolved dramatically, I believe we’re experiencing something more nuanced than obsolescence.

With AI in design, we’re coming full circle, with artificial intelligence and machine learning enhancing rather than replacing the fundamental skills we developed.

View the Episode Transcript

If you’re feeling overwhelmed by these technological advancements, consider starting with a foundational course in machine learning. Focus on understanding how machine learning works and how to query using AI. Skip the programming part because with today’s advancements, I doubt you’ll need to learn how to code.

After that, or if you didn’t feel you needed that, focus on applications specific to your engineering discipline. You will likely find the tools you use being enhanced. Combine your machine learning understanding with those tools to better understand how these tools may introduce errors into your decisions: assumptions in calculations, models being used, and data cleanliness to name a few.

You know what factors affect your decisions. So, continue to verify the answers you get. Avoid blindly accepting answers by applying critical thinking.

Three areas showing particular promise are:

• topology optimization (using AI and generative algorithms to optimize designs for weight, strength, and manufacturability)
• simulation and analysis (using surrogate models to accelerate simulations)
• material selection (finding optimal materials based on performance requirements).

I’ve also included a graphic about what you might use during which step of product development.

The essence of design engineering hasn’t changed – we’re still solving problems and creating functional products. What has changed is our toolkit, which can now handle computational heavy lifting while we focus on creative problem-solving and engineering judgment.

Everything old hasn’t gone out of style; we’ve just developed enhanced ways to accomplish our goals. AI and machine learning are giving designers unprecedented control if we’re willing to embrace these new capabilities as extensions of our engineering expertise rather than replacements for it.

Other podcast episodes you may like:

Predictive Analytics, Machine Learning, AI, and VR in Design Engineering

SOR 1045 Complex AI Reliability – Accendo Reliability Join Dianna and Fred as they discuss complex AI reliability: using AI for complex systems.

Episode Transcript

Hello, I’m Dianna Deeney and this is the Quality During Design podcast. I was scrolling through social media admit it, you do it too and I came across a post or a link to an opinion article about Generation X. I’m Generation X and the post was sort of like this Generation X is angry because we grew up in one era, learning the skills that we needed in one era of humanity, and now we are leading in a completely different era and that the skills that we’ve learned earlier to apply to our jobs and workspaces just don’t apply anymore. I kind of got a little bit of a chuckle out of it and then I moved on and kept scrolling, but I kept coming back to that and thinking about that. We are in a different era, very different from the one that I grew up in, but I don’t think that the skills that I learned when I was younger and earlier in my career are displaced. I think instead, we’re coming full circle. Ai machine learning is giving us more autonomy with design. Let me explain more about this after this brief introduction.

Hello and welcome to Quality During Design, the place to use quality thinking to create products. Others love for less. I’m your host, diana Deeney. I’m a senior level quality professional and engineer with over 20 years of experience in manufacturing and design. I consult with businesses and coach individuals and how to apply quality during design to their processes. Listen in and then join us. Visit qualityduringdesigncom. Welcome back. We’re talking about engineering, design, designing products and how AI and machine learning is affecting those processes, but I view these changes as a cycle. We’re coming full circle back around to somewhere we’ve been before, except now we have more tools and power to be able to do additional things that maybe we wouldn’t have been able to do by ourselves.

A very relatable cycle that recently happened is with music. This was another post about. Why Generation X seemed so angry is because we’ve had to purchase our music over and over again. Now my consumption of music started with radio and then records. I remember having Michael Jackson’s Thriller record album my sister bought it was listening to it on our record player at home. Then we moved to cassette tapes and we had Walkmans. Then it evolved to CDs and I remember listening to my first CD. Again, my sister introduced me to that and it was Madonna in her car and it was so crystal clear and vibrant. It was amazing.

At some point I had a 200 or 300 CD Sony device where I would load all my CDs into it. I could just randomly play from my whole library of CDs that I had. That’s how many CDs I had at one point. But then I needed to digitize my library to be able to put it on an iPod. And then I went to streaming, where I didn’t need my library anymore and sometimes I had to buy the album again. And now just this last winter I got my kids a record player because that’s what they wanted. They wanted a record player. And now, just yesterday, when my youngest was digging through a closet and found my old AM FM radio and said hey, can I use this and plug it in? I wanna listen to the radio. So it just seems like all of our music, the way that we consume music, is forever changed. We all do it a little bit differently, but it’s kind of coming back full circle again. What was old is new again. What was old is still relevant and still wanted, and I kind of relate that to this iteration of design, engineering and engineering development that we’re going through.

When I was in high school I was learning Fortran and Pascal, but then when I got to college those weren’t the things anymore Now it was Visual C++. I remember in college I would wake up from dreaming about deriving equations by hand and then, as I progressed in my career, finite element analysis came to a point where it was accessible to people in industry. But it wasn’t that we could do it in house. We had to send it out. A specialist had to have the computing power, the programming capabilities and just the knowledge of the program to be able to run a simulation for finite element analysis. It was something we had to farm out. We couldn’t do it ourselves.

We think of computer-aided design software. Again, when I was in college I might be aging myself here, but part of the graphic design is we had a kit where we would manually draw out graphics of things, and CAD software has come a long way since then, not just being able to draw but also testing form, fit and function. Those started out with specialists needing to be able to do that to more accessible to a lot more people. And then the reliability software. Well, just reliability analyses used to be really difficult until computing power and then statisticians developed some software solutions for that that made it more accessible to people, packages for statistical computing and graphics. I never got into it because I didn’t want to learn a new programming language. But now, now I can use AI to program my own package or I can check someone else’s package for some of those important statistical assumptions. You know things like normality and independence, that kind of thing assumptions. You know things like normality and independence, that kind of thing.

Now, going back to the CAD and finite element analysis, now there’s AI and generative algorithms can optimize designs for weight, strength and manufacturability. That whole topology optimization. It’s a whole specific generative AI focused on finding the optimal material distribution within a given defined space. Now our CAD systems are building these kind of analyses into them. Now we no longer have to farm it out to a specialist. Same with things like simulation and analyses. Building out simulations can be a big deal. Ai can make it easier for us to define those surrogate models.

I think AI and machine learning is making these more advanced analyses more accessible for the rest of us, which is why I think it’s giving us more autonomy with design. It’s not that the skill sets that we’ve been learning or that we used to apply are no longer applicable. Now they could be a little more powerful because we have tools to help us do it. In fact, having the baseline knowledge of how these things fundamentally work and how they’re supposed to work, the kind of analyses that they’re supposed to work, the kind of analyses that we’re supposed to do, the things that are important, that’s really important inputs into using these tools to create these new outputs that we can evaluate our designs with. It’s sort of like the statistical software where, yes, you can use statistical software to create an answer and graph and plot all of these things, but you still need to understand what it is you’re looking at and you still need to verify the assumptions of your analysis to make sure that it’s accurate. The statistical software can help you do it faster and that’s how I’m seeing AI and machine learning for other design engineering applications.

On the manufacturing shop floor, computer vision for inspection has been around for a long time Now. Can we use AI to automate those tasks? Cnns or convolutional neural networks are a subset of machine learning. Can they help us process images better and do a better job at this computer vision and for inspection and do a better job at this computer vision and fore-inspection? Still on the manufacturing shop floor, predictive maintenance is a really hot topic right now Using AI to analyze sensor data, to predict equipment failures and optimize maintenance schedules. More on the design side of things, with AI we can create our own scripts to be able to do shortcuts. With a little bit of that program language that we developed, we can reapply it in a new way to make things easier a little bit faster.

But where do we even start Now? I know machine learning and AI has been around for a while. It’s not like these are brand new they just came out last year but they are being more and more integrated with the tools that we use every day. So if we’re feeling behind on all of these advancements and are not sure what to develop for our design engineering career, I think a beginning step is to take a foundational course in machine learning. The course that comes up a lot for me is in Coursera and it’s called Machine Learning by Andrew Ng and that’s spelled N-G. Machine learning is the foundation of a lot of these different techniques that we can use for design engineering.

But then, after that general baseline knowledge, we need to get a little more specific. What is it that we want to do? What kind of engineering do we want to explore that will allow us to dive deeper into specific AI applications that are going to be relevant for the work that we’re doing In the world of design engineering. I’m really seeing three areas where we use this the most. One is topology Learning how artificial intelligence and generative algorithms can optimize designs for weight, strength and manufacturability In simulation and analysis. Investigating how surrogate models and AI can accelerate the simulations and improve the accuracy of our analyses. And, with material selection, finding the optimal materials for a design based on performance requirements. We’ve been doing this for a while, but now we can do it a little easier and a little faster. Doing this for a while, but now we can do it a little easier, a little faster.

Speaking of doing things a little bit faster and easier, we can also start to automate some of our design processes. Ai can automate our repetitive design tasks, freeing us up for more creative work, and that involves some scripting, api usage and machine learning to be able to create automated workflows. So back to our theme of everything old is new. Again, I would say everything old has not gone out of style. We’ve just developed some different ways to do it. Ai and machine learning are giving us more control over design than ever, if we only reach out to take it.

So if this is new to you, then think about taking a machine learning class. That will give you a better understanding of the inner workings and how you can apply this to your work. If you’re already using these tools for your work, let us know how it’s going. Leave us a comment on the blog or, if you’re listening from your podcast player, send me a text with how it’s going and what you like about it. If you’re resistant to changes, then let me know what is the resistance about. What concerns you about adopting some AI and machine learning in your design processes? This has been a production of Deeney Enterprises. Thanks for listening.