Brandy Ryan
Ananda Arasu
It’s impossible to be in any field right now and not have AI creep into the conversation. Automotive design is no different.
So Ananda Arasu and I had a conversation with designer Ehab Kaoud, one of our AIF presenters, to go deeper into AI and automotive design.
You can listen to our conversation here and scroll down for more context and visuals.
Why explore AI in automotive design?
Curiosity and innovation, Ehab says.
Designers are always looking for the new thing. AI offers the chance to do that, to explore different ways into ideas and process.
“We believe AI will play a fundamental role in shortening the distance between having an idea and bringing them to life for creatives.”
Kaelan Richards
AI design engines
Because this is such a new area, it’s really trial and error. You can pick an engine, try it out – and if you don’t like how it works or what it’s giving you, you move on to another one.
Two of the engines Ehab has been working with are Midjourney and Viscom.
Midjourney aims to “scale, explore, and build humanist infrastructure focused on amplifying the human mind and spirit.” You can see some of their community gallery here.
Co-created (with Kaelan Richards) by former car designer Jordan Taylor, Vizcom is the natural evolution of Taylor’s Instagram community page Designerspen. Like Midjourney, Vizcom is working to resolve the burn-out that comes with repetitive creative labor.
“The idea was originally meant to explore ways to accelerate
the car design process with artificial intelligence using
generative design methods. This idea quickly outgrew the niche
of car design and started to show potential in other creative fields
such as concept art and Industrial design and continues to find ways
to impact other creative fields.”
Jordan Taylor
AI design process
We were curious about what an AI design process might look like. Ehab shared his:
- Begin with a verbal prompt, sometimes just a keyword
- AI engine generates initial designs
- Insert the image(s) you like back into the engine to create a new image
- Provide feedback via prompts – do more of this, less of this, etc.
- You can also begin with an external image (even a mix of them) if it feels right, to give the AI engine more inspiration to draw from
- When you’ve got an iteration that feels close to your concept, you can take the finished rendering and explore it more on your own, move it into Photoshop and do your refinement.
It is a responsive process. The AI usually gives you four images from your prompts or initial ideas – and usually one of them is a bit off the grid, unexpected. It’s entirely possible the creators of the AI engine do this on purpose, to explore what the AI can create and to allow for surprise and spontaneity.
Giving feedback on AI design
Feedback can be either verbal or with other images, and this is one area Ehab notes is getting better.
For example: if you put one of its iterations back into the chat and say, “I want this area to be changed,” the AI will simply erase or modify that area, leaving everything else intact.
This is one of the exciting things about this technology—the creators of the AI engine are learning from how designers are working, just as the AI itself is. And that makes for more responsive technology and design.
The more you work with a particular engine, the more coherent the imagery is. Ehab used to say that if he were to show the initial AI designs to his boss, he’d be fired on the spot.
That’s changed. Now the images reflect what the AI is learning about Ehab’s design aesthetics and process, giving him tangible iterations to build upon.
Merging art and design in AI
Ehab notes that as a designer, he is always looking for something new, different, special. That kind of inspiration can come from anywhere – architecture, fashion, nature. You aren’t limited to feeding only vehicle images into an AI engine. You can also build off a fish or cheetah, an arch or doorway, the lines of a sleek dress.
This is part of the trial-and-error process with AI, and it’s an intriguing way to push what the AI can produce for you. Because new design is often a mix of other elements, having the AI put them together in its own ways can prove exciting.
It’s as if the collaborative process of design that your brain does automatically is externalized.
The limits of AI
Ehab notes that the human brain is more capable of generating more inspired designs than AI is currently able to do, in its Perception AI wave.
But should AI reach an Autonomous AI wave, all that could change. In that stage, AI understands the nature of intellectual tasks, can operate without human direction (no prompts necessary), and can make its own decisions (it decides when the design is done).
Right now, the engine is able to look at what’s out there (understanding and interpreting sensory experiences) and give you feedback based on that. It can’t think beyond what’s currently available, as humans do. That’s why Ehab will often feed the engine images outside the usual scope of vehicle design: to inspire it to think beyond what it currently knows about vehicle design.
Stage 1: Graphic elements fed into AI generator
Stage 2: Using graphic elements in different ways
Stage 3: Graphic elements become interior designs
It’s important to remember that at this stage inspired iteration, rather than quick iteration, is still a human realm. If AI gets there, though, everything we think we know about design could change.
Teaching students about AI design
AI is no small controversy in the visual world, and perhaps even larger in the academic world. But it’s not going away.
So the question we should ask is: how do we harness the technology so that it becomes educational rather than plagiarizing or poaching? What if we treated AI as a way to tailor education based on how particular students learn?
If, say, Ehab doesn’t understand calculus very well because the language of numbers is not one he intuitively understands, he’ll struggle. But if the same material could be taught in a visual language, he stands a much better chance of really getting it.
AI in the automotive design industry
Even if they’re not enthusiastic early adopters, most designers are looking into AI engines and how they might impact design. And some OEMs have announced that they’re using AI to design wheels, for example.
The ideal would be for the industry to see AI design as an opportunity to be explored in particular ways, not a looming threat to individual designers.
When it comes to an OEM’s bottom line, AI offers real benefits. The same is true in manufacturing. Not to replace designers or engineers, but to take on the tedious and repetitive tasks that take up a lot of human time.
If we embrace and integrate AI rather than fearing and rejecting it, we can solve problems and create better designs.
If a designer’s priorities are to come up with something new, something better, to solve certain issues that a customer may have … whether the designer solves it with AI, on their own, or a combination of the two is less important than to resolve the issue.
If AI makes it easier and faster to arrive there with better answers? How could we not explore it?
The future of AI design
AI offers incredible assistance in automating repetitive and mundane tasks, allowing designers to focus more on the strategic and creative aspects of their work. As AI continues to develop, its ability to analyze user data and preferences will increase its efficiency and output.
Just think about the possibilities for AR and VR…. AI can enhance both experiences through its analysis to provide real-time feedback, improve immersion, and adapt virtual environments.
But there are ethical concerns to AI development, particular in responsible design practices. Algorithmic bias, data privacy, and transparency can have huge impacts on AI systems. Developers need to ensure that AI systems are fair, inclusive, and aligned with humanist values.
Stay tuned for more on relevant issues like AI. You can also check out our Autodesk Automotive on LinkedIn, subscribe to our Alias and VRED YouTube channels, and follow us on Facebook.
If you’re not already an Alias or VRED user or would like to expand your skills at home, get started by downloading the new Alias & VRED 2024 Learning Editions:
For Alias Learning Edition, click here.
For VRED Learning Edition, click here.
