In its brief existence, generative artificial intelligence (AI) has both impressed and shocked due to its ability to perform tasks — not perfectly, but pretty darned well — that only a few months ago most of us had deemed too requiring of human knowledge or creative skills for AI systems to automate. Today, nearly everyone — from rank-and-file employees to managers and executives — at almost every organization in the world is trying to figure out where and how to effectively and safely apply the technology.
Two of the more common ways companies are currently using generative AI is to support their marketing and customer-response efforts. Examples include employing AI art generators like DALL-E 2 (or a slew of other tools) to assist their advertising groups or employing GPT-3-powered bots to help service reps craft responses to customer requests and other digital interactions (e.g., social media posts, product reviews).
When it comes to the latter, due to limitations of the technology (i.e., “hallucinations” and other incorrect or weird output), most organizations are not using generative AI tools to 100% automate customer interactions and responses. Rather, a human customer service representative typically must first approve and, as needed, edit any generated content before it reaches the customer.
These are some of the more common, straightforward enterprise uses for generative AI at this time. That said, organizations will want to consider how to use generative AI integrated within their key applications and business processes. This means in a manner that fits their company and brand image, character, and “feel.” Organizations should also examine how the technology can support more strategic uses. Take Audi, for instance.
Generative AI Design & Modeling
Audi has developed a generative AI application to assist its designers with creating new wheel designs for its cars. The application, called “FelGAN,” was developed in-house at Audi. FelGAN combines the German word “Felge” (“rim”) and “GAN,” which is short for “generative adversarial network.” A GAN is a form of deep learning architecture in which two neural net algorithms compete during training, eventually becoming able to generate new data with the same characteristics as the data they are trained on. For example, a GAN trained on a large set of photos of people’s faces can learn to generate highly realistic-looking photos of faces that are fictitious.
At first, you may think, “well, FelGAN is just another generic generative AI design tool.” But actually, it’s considerably more than that.
Rims & Brand Recognition
Car rims represent a very distinct — and highly visible — extension of an automaker’s brand. And there is a particular “look and feel” to car wheels that tends to stick in customers’ minds. Right after high school, I had an old 1969 Pontiac Firebird. Now there are many things I absolutely loved about that car. But the one that stands out brand-wise to this day is the wheels: it had the coolest wire-rim hubcaps, complete with a red center hub that featured in silver letters “PMD” for Pontiac Motor Division.
A New Perspective
FelGAN is intended to give Audi designers a fresh perspective on wheel design. It can generate an unlimited number of designs, including high-resolution photo-like images. Designers can take and modify these images or use them as source of inspiration for their own designs. The tool lets them experiment with shapes, colors, structure, and surfaces — all in real time. As shown in Figure 1, designers can view new design images at scale, providing a good perspective of what the wheel will actually look like.
FelGAN assigns a numeric value to each wheel design it creates. Designers can use these values to reproduce the designs as they are or utilize them as a foundation for creating new designs. For example, by submitting images of their own wheel designs to the tool, they can instruct it to generate new designs — including by combining various elements from multiple AI-generated designs with aspects of existing wheel designs. In effect, this ability to incorporate aspects of existing, established designs in a targeted manner lets designers create new wheel designs that are not just innovative or pleasing to the eye, but which also adhere to Audi’s established brand and image.
AI Design Generation Meets Physical Modeling
Designers can take their virtual wheel design and feed it into a high-end milling machine in order to turn it into a aluminum or plastic model (see Figure 2). This gives them an actual, full-size prototype to further experiment with. (As an aside, integration of generative AI tools with machinery like milling machines and 3D printers looks like a potential killer application of the technology.)
Basically, FelGAN provides an advanced, iterative design tool for Audi’s rim design team. Users can brainstorm and collaborate on new ideas by creating and sharing various wheel designs with other groups (e.g., marketing, engineering) to gain feedback, while integration with machine tools further optimizes the wheel design process by giving the design team a quick way to bridge the digital and physical design worlds.
Generative AI systems have exploded on the scene, and companies want to utilize them for business. Currently, the more common enterprise uses of the technology include for automating design, particularly in marketing and advertising, and for providing semiautomated responses to customer requests. That said, some companies are employing generative AI to develop cutting-edge applications with the potential to give them a leg-up on competitors. The bottom line, however, is that no matter how organizations seek to apply the technology, it should be done in a way that aligns with the company’s brand, style, and image.
Finally, I’d like to get your opinion on using generative AI in the enterprise. What killer business applications do you see arising from use of the technology? As always, your comments will be held in strict confidence. You can email me at email@example.com or call +1 510 356 7299 with your comments.