The results may look fascinating, but what’s cooler is that this comes from OpenAI’s DALL-E 2, founded by Elon Musk. So in a way, credit for this Apple Car goes to Tesla’s Elon Musk?! Mmm??
Meet the Apple Car, from the mind of an AI. Designed by Dall-E 2 based on a text prompt from designer, educator, and YouTuber John Mauriello, this Apple Car is fascinating for two prime reasons – the car’s design itself, but more importantly, the underlying AI technology that ended up creating the car. The genesis for this idea came from Marques Brownlee’s own efforts with DALL-E 2. In a YouTube video, Brownlee demonstrated how simply typing the words “Apple Car” resulted in a car that looked like the apple fruit. This became a starting point for Mauriello, who instead, decided to tweak the prompt a little to get more specialized results. Mauriello told the AI to design a “Minimalist Sportscar inspired by a MacBook and a Magic Mouse, built out of aluminum and glass”, while also specifying it to design something in the style of Apple’s former design head, Jony Ive. The AI churned up multiple images using those text-based cues as a reference point. It correctly understood terms like ‘sportscar’, ‘MacBook’, ‘Magic Mouse’, ‘Aluminum’, and ‘Jony Ive’. To his credit, Mauriello didn’t mention the word ‘Apple’ anywhere in his prompt, given its close association with the fruit. Eventually, it grasped the essence of those terms and designed something that made sense, rather than mashing stuff together (which is what other AI diffusion engines tend to do). The car looks quite stunning to say the least, and has a sleek, curvilinear quality to it, with a machined aluminum body and zero breaks in the surface. It’s as if Jony Ive teamed up with Jaguar or Aston Martin to bring this car to life!
The AI Engine works by creating variations based on the same prompt. This results in images that are connected by theme, but aren’t essentially the same, making the AI a great exploratory and ideation tool. There are some parts where the AI ends up making flaws, like the fact that the car on top has front wheels that aren’t touching the ground, or a few images later, where the car doesn’t have a windshield or even windows. Symmetry isn’t a strong suit for the AI too. Practically these may be industrial design drawbacks, but conceptually, the AI absolutely nailed the brief. If you look carefully at the image above, the AI even put an easter egg into the picture in the form of a warped Apple Watch wallpaper (or is it a HomePod animation?) on the rear of the car!
This so-called Apple Car, despite its minor flaws, looks absolutely stunning. Unlike other Apple Car concepts that are either too outrageous-looking or too boring and ‘car-like’, this AI-generated sportscar is sheer perfection. It’s sleek, aerodynamic, and looks just about as luxurious as something Apple itself would put out. It also helps that the AI also used wonderful dramatic chiaroscuro-style lighting to bring out the car’s surfaces! I’m not an Apple guy but I’d buy that.
An industrial design professor and YouTuber, John Mauriello has spent a significant amount of time exploring AI-based generative design tools. He’s also made an incredibly comprehensive YouTube video about Artificial Intelligence in the context of design and human creativity that’s a must-watch in my book. DALL-E 2 is currently open only to a restricted group of people, although you can sign up to join their waitlist. If anyone from OpenAI is reading this, help me out? I’ve been waiting for my invite since April!
Image Credits: John Mauriello and Dall-E 2
