A Q&A with AI Proponent Juan Carlos Noguera

A Q&A with AI Proponent Juan Carlos Noguera

Spotlight articles shine a light on designers and design materials we admire. Our founder and principal designer Rebeccah Pailes-Friedman has met many wonderful designers in her time in the industry, and in our Spotlight interviews we ask them about their work, their design journey, and what inspires them. In this interview we spoke with Juan Carlos Noguera, an industrial designer and design educator who focuses on creating holistic design solutions. He holds a masters in industrial design from the Rhode Island School of Design and a bachelor’s in industrial design from Universidad Rafael Landivar in Guatemala city. He is well known for being the product design director at Voxel8, where he pioneered the development of the first 3D electronics printer, and he is a professor of industrial design at the Rochester Institute of Technology, where he brings AI into the classroom. We asked him about how AI is changing design education, how to incorporate AI into the design process, and how he teaches students to use AI effectively and responsibly.

Juan Carlos Noguera
Photo courtesy of Juan Carlos Noguera.

Q: Design has been revolutionized in the last decade by the emergence of sophisticated digital tools, the latest of which include powerful AI generators that we have barely begun to regulate. How have you seen design education changing in light of these developments?

A: I think the first knee-jerk reaction has been lots of fear from people, from faculty, especially because every time you read about AI and education in the same sentence, you hear about cheating and plagiarism and all of the very valid gray areas of AI. It’s a valid fear that people have. I think there are still a lot of things that need to be resolved regarding the regulation of intellectual property. These are ongoing conversations that I think we should definitely keep in mind. I’m a person who’s very excited about using AI in my classes. My personal perspective is that, if we are transparent about teaching incorporating AI into the projects that we teach, if we show students what AI can do really well, what the shortcomings of using AI in your process are, how your process can suffer from using AI — I think students will organically use it appropriately, because they want to reap the greatest benefit. If we give way to the hand wringing and fear and start prohibiting it in our classes, it starts becoming a problem. 

I’ve had a chance to do things like faculty workshops here at RIT, where I try to start to dispel some of the fear around it. These systems are not black magic. There are mathematical models: they do these certain things really well, and they don’t do these other things very well. I start getting people to jump on board and feel a little bit more excited. So, I have seen a big sea change toward positivity from some people, but there are still a lot of fears. I think that’s the biggest battle right now. But let’s get people to embrace it because, if we don’t, students will use it anyway, and they’re not going to see the downsides if we don’t point them out.

Q: The role of AI tools, especially image generators, is not yet standardized in the industry. How do you see AI being incorporated into design education, and how do you see it being incorporated into the design process more broadly?

A:  It’s always hard to visualize the future. I’m trying to think what may happen when the novelty wears off, at which point do we start seeing the net benefit of these tools existing? So, for education it means a few things. One is curricular change. We start rethinking what types of assignments get put into courses. Mechanical tasks that we used to ask students to do and that can easily be replaced by AI will get replaced by AI, whether you want it or not. So how do you get that learning out of more project-based work where students engage more in critical thinking. I think it’s a good thing, it also forces us as faculty to rethink things that were easy before. It’s going to force us to make some adjustments. Having faculty become experts in the area is hard, not everybody uses technology at that level. Some faculty are more reticent to start using it or to accept that it could be a tool.

Educational institutions are ships that are hard to steer because they’ve been going in one direction for a long time, and some people don’t want things to change. We’re a bit of a tribe as designers. You have to go through a right of passage of doing this one thing. What if that thing is no longer part of the process? We were just having a discussion at RIT.  We’ve been using this AI tool called Vizcom a lot in class, and it’s basically jumping over the entire need for my students to get good at marker renderings. Now, if they have a great line drawing of their idea they pump it through Vizcom and get a good enough representation of the thing that is as good as their marker rendering would have been. So why would you learn that manual skill? There are a lot of moving parts to this, which is why I understand the frustration.

On the other hand, in industry, once that novelty wears off you’re going to start seeing these AI tools being built into things you already use. It’s going to make the tools you’re used to smarter. When you go into your CAD program, AI tools are going to be there to help you with tasks that, right now, you might consider tedious. I think Adobe in particular is really a great job at incorporating AI into their offerings. It’s naturally built into Photoshop and Illustrator, and they’re going to slowly expand on this. They’re not taking these tools and saying, Here’s AI added on top of them. They’re saying, Hey, here’s the fill tool you’ve been using for a while, we’ve made it smarter. I think that approach is very future-proof in particular. They’re calling it generative. It’s just an offshoot of their previous version. They’re doing a good job at softening that landing for people, saying, Now Photoshop does one more thing. As opposed to the crash people have when they open Chat GPT and get this ominous feeling of not knowing what is chatting back. People have a hard time getting over that. That feeling isn’t there with Adobe products. I think we will see more of that and less of the inaccessible version.

Q: It can be challenging to use AI tools efficiently and effectively. At what point(s) in your design process do you find AI tools most valuable?

A: I have found that at least some of the generative tools I’ve been using so far in my practice, and also in class with my students, are really great during the ideation phase in a number of ways. Image generation software like Midjourney and Dall-E have been great at helping teams communicate. I had a chance to do a project with artisans from my home country, Guatemala. They do bronze castings. They typically work with a designer, and the designer is up here and they’re down here, just making the thing. It feels like a very top-down relationship. We recently jumped on a Zoom call and used Dall-E to communicate visual ideas. We went back and forth and just used that as a communication tool. It really helped level the conversation field. Dall-E is almost like an automated journey mapping or storyboarding tool, so it’s great for teams. I also think image generators have great potential when a designer is just hunkered down ideating. It’s an endless firehose of stuff. And that could be good or bad, depending. But let’s say you’re tasked with designing a plywood chair, right? It’s almost like having the key to the plywood chair museum. You can ask for plywood chairs all day and you will get all these different things that are plywood chairs. Some of them will be good, some of them will be garbage. But you’ll have an endless stream of images to look at and get inspired by, which can help you work through a block. It’s like the stimulation of going to a museum. It’s more targeted, because you get to input what you want, but the mechanism is the same. So, I think it’s going to make us more efficient.

I’ve pretty much absorbed it into my workflow. It really helps me visualize specific situations as well. It’s even helped me visualize situations for my client. I’m working right now on an underwater robot model for a client. I had the CAD model for the bot and and my own rendering, and I fed it to Midjourney and said, I want this thing to be underwater on the ship hull. And Midjourney did a beautiful job of putting my object underwater and on the ship hull. That’s the kind of thing that would be very time consuming to do. All of that sort of mechanical work is not going to be client facing anyway. It’s automated now, so it’s taking away some of the more painful parts of what I do, which is what automation should do, right? I think people are worried because this is now automation for the mind, and that’s a strange concept.

Q: How do you guide students to incorporate AI into their process ethically and in a way that preserves the integrity of their ideas? What are some of the common pitfalls in AI use that you see as a design educator?

A: I think it’s important to have students use it and also kind of crash and fail with it. I like to start with very open questions and assignments when we work with AI, so that they can see that the direct output from the system is not something they can use.

First of all, they have to understand how it works. So we go ahead and talk about how a diffusion system for image generation works, how it’s been trained. The data set is owned by everyone in the world. And what you’re getting is basically reprocessed data that’s combined based on your text input, but the information is coming from unknown sources. All of these different software companies are very opaque about how their software works. This means that any output that you get from the AI, you can consider that the same way you consider the results of a Google image search. You see the results. You might find something inspiring. It’s not something you want to use directly. That’s the first thing. It’s a great way to remix things in unexpected ways and find unexpected and inspiration. But they have to remember: this is just an echo chamber of everything that’s out there in the world. If you actually want to generate something new, you can’t use that output, because that’s not how the system works. You won’t get something new, it’s just giving you information from its data set. I’ve tried to get students to use it as a way to generate rich mood boards and inspiration boards as opposed to trying to get form from it.

Prompt engineering is on the horizon of being an actual profession. Out of all of these systems I like to have students play with Chat GPT. It’s a great way to train your thought process. They need to learn how to ask things from AI systems. You can give it a lot of context and get really great results back. We did this experiment in class where I said, Okay, I’m going to write a letter or recommendation for one of you, which is a task that I do as a teacher all the time. And at first I just asked Chat GPT directly, I want a letter of recommendation for this student. The student is an industrial design student. It gave me something professional and polite, a very generic letter of recommendation that sounded like a template, it sounded robotic. It was bad, it was not something usable. On the other hand, we repeated the exercise. We started describing my relationship to one student in the class. I said, Hey, Chat GPT, my name is Juan. I’m an industrial design professor. I come from Guatemala. I’ve worked on this and that before, these are my interests. You describe them in a couple sentences. And I said, I met the student last semester. They took my drawing class. They did this. The prompt was probably as long as the letter that I needed to write but I wasn’t trying to word it well. I had grammatical errors. I was being very casual about just telling it a lot of information. The language model is great at processing that. And then after that I just said, Knowing all that, please write a letter of recommendation for this person. And that letter was amazing. It was great. I would not guess it was written by AI. I seemed connected, like I really knew the student wanted to speak to the things they did. When you do an exercise like that, the students really see the power of the right prompt. They see, okay, I need to get context, I need to think carefully about keywords and information and try to frame my thoughts. After that, they start using the tool in more creative ways.

I use the tools in class as much as possible to try to take away the shame element of using AI. I say, Use it as much as you want. You just need to tell me when you use it. You need to cite it almost as if you’re citing a source for an image or anything else you’re putting in your paper. I want to know which steps used it, and how that helped inform your design decisions. Once we get that established, they feel more at home using it. They know that they’re not cheating, it’s not plagiarism. They’re using it, they’re being transparent about what they got from it, what they couldn’t get from it, and what failed. 

Q: What reaction are you seeing from the students? I know there can be costs to using these tools.

A: Yeah, there’s grumbling on the cost, but the same way that we’ve been grumbling about the cost of software forever. The same way I grumbled about paying $100 a year for my Keyshot subscription. Or when I finally graduated from school and realized that my professional license for SolidWorks was $5,000. Students here at RIT don’t get free Adobe licenses, they have to have their own. And that’s always a problem. So, that really hasn’t changed. 

One thing I do notice that is a source of frustration is going back to curriculum and course development. They now start seeing, Okay, here’s this tool, and it does the thing that you’re teaching me how to do by hand. So why are you teaching me this again? It feels redundant to them. I think that has been the most painful part of it for my students, just trying to figure out: What is the skill set of a designer in the end? What’s desirable when I go out into the world? Should I have AI skills? I might not be getting them from the school. Do I have to get them on my own? Should they be in my CV? All these things are a source of frustration.

Q: How do you see the relationship between AI and design education developing in the near future?

A: That’s a great question that I don’t have a very solid answer for. Last semester we had about five or six different faculty forums about what’s going to happen with AI and education. Universities are wrestling with how to tackle it and these tools are popping up that are really unexpected, like Vizcom. It basically showed up a couple of weeks before we started the semester and it really made us think, Should we try to retool the entire design drawing class we’re teaching this semester? Because they’re going to use this and wonder why they need to learn how to work with Copic markers. Those are valid questions, because… is this a skill that is no longer relevant?

We need to ask those hard questions. That’s definitely on the horizon because I think this is significant enough that we’re going to need to rethink many parts of our curriculum and how we do things in education. We can’t treat it like a piece of software. We have to treat it as the sea change that it is. It’s very fast-paced, and we don’t know what other tools will pop up. They’re going to replace everyday tasks. This has already started to happen. When I went to school, I did three semesters of technical drawing. Giant board with a parallel ruler. That now feels like it was a waste of time, but that was the industry standard when I went  to school. That change happened gradually, but this is happening fast, and I think that’s the big difference. 

The school-wide policy in most schools that I’ve seen has been very nebulous. It’s mostly been about trying to give faculty a little bit of power over how to use AI in their class. The syllabi that we use now have language that says that the faculty get to say when AI is appropriate or not, or that it’s not to be used unless the faculty instructs the students to use it. This leaves the door open for usage while still trying to be cautious. institutions are trying to use caution, which is natural, right? But there should be a sense of urgency. And that really depends on who you ask. 

I would say generally that industrial designers are overall super excited. Very happy. They can see lots of applications, things that can help us. If you’re a graphic designer, you’re less excited. If you’re Illustrator, you’re definitely not very excited about this at all. A concept artist? You don’t want this at all. It really depends on your specific area of interest and also your industry. There are a few industries that might be more directly affected by this existing. If you work in an industry with a very specialized knowledge base, soft goods for example, I would be less worried. If you work in another area, maybe traditional consumer goods, plastics, it’s hard to tell.

I’m very optimistic and see great benefits using AI in my work. It frees me up for what I consider the more critical parts of designing, like trying to be user-centric, trying to be sustainable in my decisions. It’s reducing the workload of things I don’t like to do, which is what automation should do. Not everyone is that optimistic, but I see good historical precedents for this. When photography was invented, everybody thought, This is going to destroy painting. Most painters were tradespeople, because painting was a way to do naturalistic documentation. But the rise of photography spurred painting as an art form, and all these different currents, like impressionism and cubism. It was a documentation tool that was taken over by this other, more automated thing: photography.

When they invented the sewing machine, they thought, It’s going to replace the people that create clothes. And there were strikes and all this resistance. It took 30 years for it to catch on. And people realize, no, the sewing machine just makes people that make clothes more efficient. It drives up quality, it ends up being a good thing. But there’s a lot of fear. 

When we started having records for music, musicians unions got the radio station consortium of New York to play records one time and then destroy them. Because before that, music was live, right? So you have to pay the band for every play. Records took that away. So buying one record for each play was the way that they saw out of it. There’s always a clash but, in the end, all of these technologies helped make things progress. I think that’s what’s going to happen here.

Check out the rest of our Insight series to learn more about the design industry. Sign up for our newsletter and follow us on Instagram and LinkedIn for design news, multi-media recommendations, and to learn more about product design and development!