The Future of Fashion & Sustainability | 11 Fashion in The Age of AI
As we commence the final part of our series on fashion and technology in 2023, the next two issues will address the impact of Artificial Intelligence (AI) on the fashion industry. Generative AI, a form of machine learning within AI, has captured the attention of consumers and corporate boardrooms alike. Generative AI uses algorithms learned from data models to generate images, texts, videos, and audio. The introduction of OpenAI’s language model-based chatbot ChatGPT in November 2022 has spurred a frenzy amongst corporations worldwide to harness the capabilities of this new technology to modernize their businesses and gain a competitive advantage. While non-generative AI technologies are already being used in the fashion industry to manage supply chain logistics, inventory strategy, and apparel fit technology, the advent of generative AI has sparked even more interest and urgency in integrating AI into businesses. [1] A partnership between fashion brand Revolve and generative AI studio – Masion Meta is just one example of how generative AI is being used in the fashion industry today.
Businesses across the fashion supply chain are excited by the potential boost to business. Management consulting firm McKinsey & Co.’s annual generative AI report estimates a $150-275 billion bump in operating profits from the adoption of AI to the apparel, fashion, and luxury sectors in the next three to five years.[3] There are many potential use cases of generative AI on the runway to marketing campaigns to studios. AI can be incorporated into product creation, virtual clothing try-ons, supply chain management, and customer experience.
Early surveys of generative AI tool usage in the fashion industry indicate quicker, more seamless workflows, “allowing workers across fields including marketing, operations and engineering to focus on more creative and problem-solving tasks.”[2] As with other industries incorporating AI, it is too early to determine how this technology will impact worker job security. However, incorporating AI technologies over the past year has been shown to optimize tasks and augment the capabilities of the current workforce.
Up until now, brands have been hesitant to incorporate AI functionality into the design process, cautious of the possibility of replacing human creativity. Despite these concerns, the value of generative AI is in boosting rather than replacing human creativity, freeing design teams from manual tasks. Creative directors can partner with the technology to generate a wide variety of design options based on data ranging from past products, stylistic characteristics, trends, and more.
There are already tools that incorporate design generation from chatbot-like text commands. Fashion supply chain interface Cala has integrated DALL·E 2 – a text-to-image AI model developed by OpenAI to generate images and art – and a more advanced language model GPT4 to provide their customers with the ability to create detailed AI-designed clothes.[5] Users can specify product specifications such as stylistic characteristics and product categories in a text interface that subsequently produces various design ideas that match a creative directors’ vision.[4] According to Cala CEO Andrew Wyatt, Cala is seeing hundreds of new users including “several of the top 25 global fashion brands” along with independent design labels and fashion students.[6] Spanish fashion company Mango developed its own chat-based generative AI platform Lisa “to design its collections and support customer service processes.”[7] Within this process, the human touch is still required to determine the market readiness – from fit to materials to pricing before the merchandise is ready to be produced.
Companies are also exploring AI technologies to forecast upcoming trends that inform design decisions for future collections.[8] These forecasting tools detect patterns by scanning datasets that include runway show images, social media posts, and in-store/online sales data.[9] Spanish fashion brand Mango’s internal AI platform, Lisa, also used AI trend analysis to aid the company’s product creation and design process.[10] AI-based forecasting tools can derive information on regions, consumer age groups, and culture-specific trends, which helps brands make informed decisions on future collections. Trends in fashion can change quickly, so AI tools allow brands to plan for future collections with the most up-to-date data. AI-based trend forecasting could also have the unintended consequence of promoting a fast-fashion business model where brands use up-to-date forecasting data to generate shorter apparel production cycles and produce more clothes.
Despite the stated benefits of AI-centered design, companies should weigh the potential costs associated with relying on AI technology. Using generative AI tools that can turn text-based commands into creative design options may have little barrier to entry. However, early iterations of tools may produce designs that may not translate to a product ready for creation. The design process still needs a human touch to determine the viability of materials and apparel structure, among other factors.
AI-generated design technology raises concerns about intellectual property. Generative AI models create their output by learning patterns from the data fed into them. These models rely on real-world images and related stylistic features consistent with any fashion brand. There are questions regarding the copyright protection of designs that are inputted into AI models, as well as the designs that are generated from AI tools.
Approaches to the protection of AI-generated designs may vary between the U.S. and other countries. In a release from March, the U.S. Copyright Office acknowledged that “where an AI only receives a prompt, the resulting expressive work is not copyrightable,” and there have been additional legal debates since.[11] Most notably, an art piece dubbed Théâtre D’opéra Spatial by artist Matthew Allen was denied copyright by the US. Copyright Office despite “624 text prompts and input revisions” along with Adobe Photoshop use.[12] The U.S. Copyright Office determined the artwork was not a product of human creation. It is important to note that the U.S. does not offer copyright protection to fashion designers, but European law does provide protection outside of just copyright law. It is unclear whether AI-generated fashion designs will receive protections in the U.S. or elsewhere, but the use of generative AI tools in the fashion design process will continue to prompt legal discussions.
It is common for people to have questions and concerns about the type of data that is used to train AI systems. In the fashion industry, the line between inspiration versus outright copying of designs is a continuous debate in the industry, and AI-generated designs may further pose issues for large fashion companies and independent designers. Unlike other creative industries, such as the movie industry where the Writers Guild of America negotiated with movie studios to protect the work of writers, there is no such legal protection for fashion creators in the U.S. or abroad. [13]
The fashion industry is currently experimenting with the use of AI, both generative and non-generative. However, it is early to understand the true impact of AI tools on the fashion industry’s growth, productivity, and workforce. Business projections and sentiments from creators at all levels indicate a growing interest in generative AI technology, but many have yet to adopt the technology. Like any burgeoning innovation, there will be an increase in adopters over time. However, I believe that large and independent brands that can balance efficiency gains with human creativity will be the ones to succeed. In our next article, we will explore how one partnership is driving early experimentation of generative AI-based product creation.
[1] https://3dlook.ai/content-hub/artificial-intelligence-in-fashion/
[2] https://www.glossy.co/fashion/how-generative-ai-will-impact-fashion/
[3] https://www.mckinsey.com/industries/retail/our-insights/generative-ai-unlocking-the-future-of-fashion#/
[4] https://www.mckinsey.com/industries/retail/our-insights/generative-ai-unlocking-the-future-of-fashion#/
[5] https://ca.la/ai
[6] https://www.voguebusiness.com/technology/a-new-fashion-brand-is-rethinking-production-using-ai
[7] https://english.elpais.com/technology/2023-10-04/the-mango-fashion-company-developed-its-own-ai-to-design-clothing.html
[8] https://illumin.usc.edu/applications-of-technology-in-fashion-trend-forecasting/
[9] https://www.theguardian.com/technology/2023/oct/01/ai-artificial-intelligence-fashion-trend-forecasting-style#:~:text=These%20AI%20tools%20can%20detect,and%2C%20crucially%2C%20more%20quickly.
[10] https://english.elpais.com/technology/2023-10-04/the-mango-fashion-company-developed-its-own-ai-to-design-clothing.html
[11] https://www.reuters.com/legal/legalindustry/copyright-conundrum-protection-ai-works-2023-11-28/#:~:text=It%20states%20that%20in%20cases,the%20manipulated%20work%20can%20be.
[12] https://www.wired.com/story/ai-art-copyright-matthew-allen/
[13] https://www.reuters.com/legal/legalindustry/copyright-conundrum-protection-ai-works-2023-11-28/#:~:text=It%20states%20that%20in%20cases,the%20manipulated%20work%20can%20be.