Step into a world where art meets artificial intelligence, where creativity intertwines with cutting-edge technology. The realm of fashion, long the domain of innovative designers, is now witnessing a riveting revolution spurred by Generative Adversarial Networks (GANs). These powerful algorithms have sparked debates about the future of fashion- will they herald the demise of traditional fashion designers or elevate their craft to unprecedented heights? I will argue that GANs can never attain true creativity, and thus that human designers will always be needed – however, using GANs may enhance the design process. But first, what are GANs?
What are GANs?
Ian Goodfellow and his colleagues originally designed GANs in June 2014. Jason Brownlee describes GANs as deep learning “algorithmic architectures that use two neural networks, the generator (generative network) and the discriminator (adversarial network), pitting one against the other to generate new, synthetic instances of data that can be used as real data”.
In the case of the generation of designs, Jason Brownlee explains that the generator creates very realistic samples that closely match the programmed dataset and so attempts to “deceive” the other network into believing its work is authentic. The discriminator is then programmed to attempt to tell the difference between original and fake artwork and provide feedback to the generator on improvement. Both networks gradually improve their outputs through the process of competition and feedback. At the end of the training, GANs produce new fake images that look like real images.
How are GANs used in the Fashion Industry?
GANs can be used in the fashion industry for creative purposes in several ways. For instance, the German e-commerce giant Zalando is investigating how GANs may be used to create pictures of fashion models wearing customer-selected clothing products, allowing their shoppers to visualise how the separate pieces of an ensemble might fit together in new and exciting ways. According to Henry Ma, customers may also utilise augmented reality technology to preview how a particular garment might appear on an AI fashion model, or even on themselves, before purchasing it by using the camera on their smartphone or making use of virtual reality glasses or even augmented reality contact lenses. The only way to address this issue previously would have been to photograph a model wearing each combination of apparel multiple times – an expensive undertaking. In addition, due to the effects of the COVID-19 pandemic, people would not want to try on clothes that others might have touched. Hence, demand for online shopping has grown, and GANs help facilitate this process.
Will Knight notes that Amazon, the giant e-commerce company, is advancing that kind of algorithmic strategy even more. A group of Israel-based Amazon researchers created machine-learning algorithms that determine if a given style qualifies as fashionable by looking at picture descriptions. The algorithm may offer style criticisms or correction suggestions as well.
These illustrate how GANs are already transforming the online fashion industry. However, considering how creative GANs are in their use in the fashion industry, can we now say they are ready to replace human fashion designers?
Can GANs replace Fashion designers and Models?
One of the crucial things that human designers contribute to fashion is creativity. If GANs are to replace human designers, they will need to be creative. Now, as we have seen, GANs are able to create new ideas. They are even able to do this autonomously, as they change their outputs through their training in ways that programmers could not imagine. However, there is more to creativity than just being autonomous in creating new ideas.
Margaret Boden suggests that creativity involves generating ideas that are new, surprising and valuable. She further argued that such a process has to be done autonomously, intentionally, consciously, and with emotion to consider such work to be born out of genuine creativity. This suggests that there is more to creativity than just being autonomous.
In my research on AI creativity, I add to Boden’s account by claiming that creativity requires that our ideas are inspired by our first-person experience of the world. Creativity is about a matter of sharing our experience of the world in our own unique way and style, and if something does not have first-person experience, it has nothing to share. This is a real problem for the claim that GANs are creative. This is because, irrespective of the fact that GANs can generate new, surprising and valuable ideas, they do not experience the world from a first-person perspective, but rather from their programmer’s perspective.
During training, GANs learn to capture patterns and features from the data they are trained on. This process allows them to generate new content that can be novel, surprising, and valuable. While GANs do not have a subjective or first-person perspective like humans, they form representations of data that might be considered a unique perspective from the data they have learned. However, this perspective is derived from statistical patterns and relationships rather than an experiential or first-person perspective, as humans have. Therefore, while GANs can create new content, their “perspective” is fundamentally different from the human experience, which is why some argue that they might not exhibit creativity in the same way humans do.
In essence, if one does not have that experience, then one does not have the tools to be creative. This means that GANs are not creative in the same way humans are. This ultimately implies that GANs are not creative enough to replace fashion designers. One might raise an argument questioning whether creativity is something that is so important in models. My take is that creativity in models like GANs is increasingly seen as an advantageous trait. While these models primarily aim for accuracy and efficiency, introducing elements of creativity can enhance their problem-solving abilities. Creative models can generate novel solutions, explore unconventional approaches, and offer diverse perspectives that might not have been apparent through conventional algorithms.
Incorporating creativity can also foster innovation in various domains, leading to unique and valuable outcomes. Therefore, meaning that creativity is important for models. My point is that the manifestation of creativity in models differs from human creativity. This is because GANs are more about generating novel and valuable outputs based on learned patterns and data, not from their first-person experience.
Can GANs help human designers?
However, rather than worrying about the potential for GANs to replace human designers, we should try to utilise them to enhance our own creativity. The need for self-expression in humans is profound, and this need will persist. Therefore, enhancing human creativity is where the actual potential of creative, generative technology lies. To better convey their feelings, people want to delve deeper inside of themselves. They aspire to improve as artists, but only to the extent that their own distinctive artistic expression reaches new depths.
I think GANs can work with fashion designers, to help them discover their true designs. This is the same with other technological developments, such as electric pianos, Photoshop, digital audio workstations, and Canva, which are naturally used to increase human expression. Systems using generative AI are not any different. In short, because it acts as a creative partner, generative AI is best suited to serve as a personalised guide to help us uncover hidden aspects of our own creative expression. For a very long time, labour-intensive tasks have been accelerated by technology. The fact that these were not the main objectives of computing advancement may have slipped our minds. The possibilities are much bigger when it comes to generative AI. The opportunity is here for us to push the boundaries of human inventiveness. Instead of asking whether GANs will take over the fashion industry, we should focus on grabbing this chance and using it to have the greatest impact on the fashion industry.
Comments