This article is a transcription of the #15 episode of IDology - the industrial design podcast by Mindsailors. You can watch the entire episode on YouTube or listen to the audio version on Spotify, Apple Podcasts or Google Podcasts.

In the #15 episode, our company’s COO, Voytek Holysz, sat down with the co-founder of industrial design company Mindsailors and senior designer himself, Rafał Piłat, and Anita Rogoża, a designer and researcher at Mindsailors, to talk about using Generative AI in product development. Our experts discuss the potential of AI, exploring how it can support the development of new products. You will learn whether and how AI can be helpful in practical application, from the idea stage to the project realization, and what opportunities it opens up for creators and designers.

Voytek Holysz: Anita, let’s jump right into it. Do you use AI tools in your work?

Anita Rogoża: I do, I do. And I use them to a limited extent, to be honest. I find them most useful in the pre-design phase: the creative, the inspiring or researching phase. For example, while I do research, I can jump towards more specific fields more easily while I am using ChatGPT, because with that I can quite fast collect a lot of samples about different subjects in my research. So that’s quite useful. I also use AI tools while inspiring myself. For example, in Midjourney I can create some wild mashups. So, let’s say that we are creating a vacuum cleaner, Apple-style vacuum cleaner. I can create some inspiration with fast mashups. They won’t be industrial, great designs, but they can lead me towards making some new and interesting forms by myself.

Voytek: They are meant to be inspiration, right?

Anita: Yeah, right.

Voytek: Rafał, how about you?

Rafał Piłat: Yeah, remixes are cool. Although I did get funny results trying to get a calculator designed by Apple. So I remixed the calculator with a text prompt, and all I got was an apple-shaped calculator as a result.

Voytek: Okay.

Rafał: Yeah, but overall, this is a great tool, because obviously, it can generate pretty quickly a large number of inspirations. So, I treat this as a starting point, when looking for an interesting shape or something like this. Well, I also use ChatGPT which has many use cases for me as well. And I also use ClaudeAI, which is a competitor to ChatGPT. What I like about it is that you can actually upload a pretty large amount of your starting material.

Anita: Cool!

Rafał: So this is cool, because you can work on a specific subject. So you can upload that part of a book and just work on the text of the book, or just report and just work it out from them. So this is a pretty good tool as well. In terms of creative tools, I use Stable Diffusion. I use it a lot for beefing up our 3D renders.

Anita: Oh, yeah! That’s a cool case.

Rafał: Especially those environmental ones, which we try to reach for photo-great, let’s say, material. We can then quickly generate highly realistic/photorealistic pictures of people using our designs in real environments, using Stable Diffusion. And this raises the level of the fidelity of the render pretty high. And what else? Oh yeah! I have never been good at sketching. That was never my forte and that’s why I use Stable Diffusion to sometimes generate sketches from my 3D models. Believe it or not, this is sometimes required to work the other way around. So this is the tool that I use because it saves me a lot of time.

Voytek: So you start with making the 3D model and then reverse engineer it into a sketch?

Rafał: Yes.

Voytek: That’s interesting.

Anita: Yeah, that’s very helpful while making the marketing materials and all of the little graphic work we have to do sometimes. That’s a lot of help, I guess.

Rafał: Yeah, we had a case recently, where we had to generate a pretty significant amount of sketches. And it turned out that the best way and the fastest way was actually to generate those sketches from screenshots from SolidWorks. So that was a very good use case for us, where we could speed up the process significantly.

Voytek: That’s amazing how, apart from the obvious methods of using AI tools like generating images, or inspiration, or assisting research, it can be simply used to speed up a process. Like, what you just mentioned - there are also tools for, for example, increasing the frame rate of a video when you are rendering animations for your clients. Then you can just render them at half speed and that half resolution and you can then scale it up which is a lot faster and can still generate great results.

Rafał: That’s true.

Voytek: And for research, or marketing, there are really cool tools for impersonating or trying to fake audience tests. Like your personas, where you can do tests with a specific audience, which is AI generated and which reacts in a specific type of roleplay, sort of simulation. So that’s a lot of assisting sort of tools. What about— Because you started with the phase which is before actual design work; you have mentioned research, you have mentioned marketing, you have mentioned inspiration and mood boards. But once you get the inspiration you need, once you get all the creative juices flowing, how much current state of AI is of use to you as a product designer?

Anita: Not much because the CAD type of designing stuff or CAD grade modeling requires a lot of technical knowledge. That technical knowledge has to be put into fresh, never existing, never been seen before forms. So I think this is the kind of struggle for AI at this moment: we are fitting it with what’s been already done. And sometimes we are working with products that have never existed. For example, new types of medical stuff that are not very intuitive to interpret while you look at them. So I think this is something in the making. I think it will happen in the future, but it’s still in the developing phase, I believe.

Rafał: Yeah. I recently tried to describe a picture of a medical device that we designed some time ago and the result was not that great. Well, I still have to try the newest ChatGPT, because they actually are claiming that it’s really good at describing what it does see and at interpreting this—

Voytek: What do you mean that the result that you received wasn’t good?

Rafał: Well, imagine if I showed you a product, which already exists, but you still wouldn’t be able to know its function just based on its look. That’s why it’s really hard to describe something - it’s even hard for a human and not to mention an AI.

Voytek: Okay. So it’s like when a real person grabs something and says, “What is it? What does it do?” That’s the same problem for AI.

Rafał: That’s right. It deals with something it has never seen. And is having simply a hard time trying to even guess what it is because the guess was far from accurate.

Voytek: Well, this must be especially useless when it comes to developing things that are novel, new.

Rafał: Yeah. Well the other way around when you expect highly imaginative work, it’s still something that I see that the AI is struggling with. I guess it’s really hard to generate something just from a text prompt. The best results that I got was mixing a text prompt with an image and then trying to reach some kind of level of remixing and trying to use it as a starting point for evolution for this artwork. Yeah, it’s a tool. That’s what it is.

Anita: Exactly. We are always falling into that chase with those tools. We kind of know what we want to reach, what we want to have as an end result. And it’s quite tough sometimes to talk it through with AI with prompts and stuff to make it happen.

Rafał: Sometimes it's hard to talk to a person.

Anita: Yeah, sometimes it's hard to talk to the person. Yeah.

Voytek: It would be nice if we had prompts for people. Like you could test different prompts.

Anita: “Understand me.”

Voytek: —to better communicate with a person. How important is  it, you think, for designers to learn to use prompts correctly nowadays? Is it important or not so much?

Anita: Language has always been important.

Rafał: Yeah. It’s a skill because you might as well not have a skill for 3D modeling and you will simply not be good at it. So I guess it works the same with prompting AI to generate something - if you cannot focus your thoughts and verbalize your idea, then you won’t be able to describe it not only to AI but to another person as well. I guess part of our work, and it’s rather a huge part, is trying to get from our clients the information that we need to start working on any kind of project. I am really wondering whether we could, in some time, use AI as an assistant to interview our clients to get the most out of them in terms of generating a brief for our project.

Voytek: Or to answer those difficult, long emails.

Anita: Yes.

Rafał: Oh, yes!

Voytek:In a calmer manner.

Rafał: I see this as a future challenge for AI developers.

Anita: Maybe it will be like that: on the one side we will be using AI, on the other side the client will be using AI and humans will be just exchanging the outcomes of those engines.

Rafał: We will be just designing with AI.

Voytek: Yeah, both sides will give the prompt. Just get it done! Just get it done! And the AIs will do the battling for us.

Rafał: And we can—

Voytek: —Just do our work.

Anita: Just do our work, finally.

Voytek: And so, from what you are saying, generative AI or any sort of AI, for example using prompts, is a necessity. So you generally should learn its language and should be able to use it to your advantage. It’s not like a gimmick or something for fun, once in a while “Maybe I will use it.” It’s something that’s here to stay in your work. Yeah?

Anita: I believe so. We can see that also in the tools we are using because for example, Photoshop, the huge software that is very important for graphic design, has already implemented AI tools in itself.

Rafał: Sure! It’s called generative fill. It’s been in beta for some time. And now, it’s in the latest version of Photoshop. Well, in terms of functionality, you can simply mark part of the image and ask Photoshop to generate something. And you need to describe it with your words because this is a text prompt that you are actually using in a graphic editor, so that’s part of it. But in terms of general functionality, it doesn’t give the user much control because in comparison, for instance, to Stable Diffusion, there is very little control over what is generated. So actually, the effects are very random from something completely of what you wanted to achieve to some good results.

But overall, I guess, this is a novelty for the moment. Adobe has presented just days ago something that they are working on, which changes the Photoshop functionality. Well, it changes the general approach on working, on the workflow in Photoshop. They try to get rid of the layers and introduce object working. So the AI will be able to simply select an object that it can see in the picture and replace it with something else. And of course, the demonstration was perfect and it looked very good. But this is something in the future. So I guess once this podcast is published, maybe we will be able to test it in beta.

Voytek: Okay. So if this is the present: tools that assist your creative work or can help you make it easier, or give me more time for your actual work while saving your time somewhere else, what is the future of those tools at your work?

Anita: I think they can become more dedicated to the needs of the user. For example, I know that there’s one architect studio, Zaha Hadid Architects, I believe. I think they are using AI to create concepts because they taught it their style.

Voytek: So they fed it with their designs.

Anita: Yeah, with their designs. So I think that might be quite a useful case in the future.

Rafał: Yeah, they actually claim that they use about 10% or 15% of the designs that they tend to generate, and they use it for their actual work. So I guess this is a good example that this tool is useful.

Voytek: What do you guys think personally about it? Like, you go to a design company and you know that they sort of skipped 30% or 40% of their work by generating something based on their previous design, so not necessarily providing you with an original 100% handmade design. Is this something that will change the value of work in studios you have mentioned?

Anita: I don’t think so because in their case, you kind of come for their significant style. You want that styling, you want that look, and I think you want a project that will continue that styling.

Rafał: Recently, I have found out that I have some style as well.

Anita: Finally you have some style! <laughter>

Rafał: So I am learning new things. <laughter>

Anita: You are teaching AI.

Voytek: Now you are learning your style.

Rafał: That’s right. So this is an interesting part. The answer to your question is, “Yeah! We are using our knowledge and our past experience when we are designing something new”. So obviously, we will incorporate our past work into something that we create. And as Anita said, when you come to a studio for a specific effect, you might be looking for a specific style as well. So, I guess that’s the answer to this question.

Voytek: Okay. Just to sum up for a closing argument, do you guys think that every designer should learn the skill of the tools of using the prompts to be at the frontal wave of new AI tools? Or should studios hire AI experts that will be leading the innovation in their companies? What’s the future? And how do you think that AI tools are going to change your work significantly, or not really?

Anita: You still need to know what you want to achieve and still need to know what effect you are aiming for. So, I think that will speed up work, be a new inspiration, new tools, but I don’t think it’s going to be—

Rafał: Revolution.

Anita: Revolution. That’s right.

Rafał: I guess, I am really actively waiting for a client that will come to us. And they will come with generated concepts using AI. And we will have really smooth work because this is a starting point for us, we can then continue working on the specific early concepts from our client. That would be something excellent because he or she would already have the idea visualized.

Anita Rogoża is an experienced researcher & designer at Mindsailors. Her priority is on making functional designs that are both user and environment friendly.

Rafał Piłat is an entrepreneur and co-founder of Mindsailors, an awarded industrial design company, with over 18 years of experience as a designer himself.

Voytek Holysz is the COO of Mindsailors with 16 years of experience in running a business in creative B2B services, marketing, sales and video production.