The Future of Generative AI in Fashion
Photo by Google DeepMind on Unsplash
Like any industry, fashion businesses seek improvements that generate a greater return on investment. Unlike other industries, the fashion supply chain is complex for a product with consumer expectations of relatively inexpensive costs in a highly competitive market. The improvements should deliver faster speed to market, more significant innovation, greater understanding of the market and consumer behaviour, greater operational efficiency reducing overheads, and higher quality products, to enable fashion brands and retailers to generate better returns to the bottom line.
However, the Fashion Industry has a comparatively high negative impact on the environment - the Fashion Industry is estimated at 2% of global GDP, yet contributes up to 10% of global Greenhouse Gas Emissions and is the 2nd largest consumer of water in the world.
Several industry initiatives have aimed to improve the sustainability of the fashion supply chain and products, though they have yet to be as successful as hoped. The recent turning point is that federal and state governments will soon impose legislation and externally defined targets for science-based measurements as evidence of sustainability improvements for the fashion industry.
New York state bill, the Fashion Act, will affect all companies, and their supply chain processes, associated with operations generating over $100M in annual revenues. This covers all the usual global brands.
The proposed bill has three main elements;
1. Binding targets: targets defined by the governing body and must be published by each qualifying company
2. Complete supply chain mapping: full mapping of the supply chain associated with operations in New York state, with the objective of traceability of products. The initial requirement is: “a minimum of 50% of suppliers by volume across all tiers of production shall be mapped”.
3. Social and Environmental impact disclosure: regular publication and comparison to binding targets which an objective and independent body will assess
The proposed punishment for brands and retailers that are not compliant is 2% of annual revenues.
The E.U. Green Deal was approved in 2020 as a set of proposals to improve policies in line with greenhouse gas emission targets. One of these is the Sustainable Products Initiative (SPI), a new regulation to enhance E.U. products’ circularity, energy performance, and other environmental sustainability aspects published in March 2022. The SPI was set to have a public consultation on priority product categories in 2022, but this has been delayed. Ultimately, the SPI is set to affect brands starting in 2024. However, brands will need to take action in 2023 to avoid penalties.
The Digital Product Passport (DPP) initiative is part of the Ecodesign for Sustainable Products Regulation (ESPR) and one of the key actions under the EU’s Circular Economy Action Plan (CEAP).
The DPP is key to the EU’s transition to a circular economy and will provide information about products’ environmental sustainability. It aims to improve traceability and transparency along the entire value chain of a product and to improve the management and sharing of product-related data critical to ensuring their sustainable use, prolonged life, and circularity.
The DPP will help consumers and businesses make informed choices when purchasing products and help public authorities better perform checks and controls.
The critical aspect of these legislations is that they will not be self-governed by the fashion industry, so it will not be a case of brands and retailers marking their homework. There have been many self-governed initiatives – just one of the many reasons why a fashion industry-wide standard of data gathering and systems collaboration has not been achieved to date. The independent assessment and enforcement will provide a greater probability of success.
The legislature will define all standards for compliance with the bill, which is a significant change from past industry-managed initiatives.
On the 1st of June 2023, the EU27 agreed unanimously on bringing forward the legislation for DPP and a ban on the destruction of unsold textiles and electronics. The EU parliament can pass this into law as quickly as logistics allow, and today, the 15th of June 2023, the Committee on the Environment, Public Health and Food Safety (ENVI) voted in favour, which is a key next step towards adoption ahead of the plenary vote at Parliament in July.
These legislations aim to introduce transparency and traceability into fashion supply chains. By definition, when this is completed, the supply chains will generate a massive volume of data – the precise requirement for effective machine learning (ML) and subsequent Generative AI.
The initial process mapping and data integration required will require much effort. Still, they will then ‘open the door’ to the continuous sustainability improvements that are the aim of the legislation. Generative AI will become a powerful tool in this continual improvement process. Let’s look at the Generative AI examples in the Fashion Industry today, and then we’ll discuss the possibilities in the future.
Generative design: G-Star used generative AI to design a denim collection inspired by natural phenomena such as lava flows and rock formations.
Unique design and styling are rapidly generated for review and have proven powerful design results. However, it has been challenging and time-consuming for the pattern technicians to interpret the design images and create the physical garments.
Generative modelling: Levi’s used generative AI to create e-commerce models that matched the style and fit of their products.
A mix of human and virtual models reflecting their consumers’ size, body type, age and skin type helps to diversify the human models, use AI-generated models to expand the number of models per product and reduce costs and negative environmental impact.
Generative content;
Marketing images and Chatbots: Revolve and Prada Beauty used generative AI to create marketing imagery tailored to different audiences and platforms. Kering and Zalando used generative AI to create customer-facing chatbots with product recommendations and styling advice.
Product descriptions: Adore Me used generative AI to write optimised product descriptions for SEO and conversion.
The examples above provide minor improvements to speed to market and sustainability but not the significant advances the fashion industry requires.
There are many use cases for generative AI in 3D modelling – it’s well established in engineering 3D CAD – but maybe a little longer before it becomes a standard feature in 3D-DPC applications.
However, if we widen our thinking to the massive volume of variations of products and data throughout the fashion supply chain, we see fantastic opportunities. What if, with the input of a Seasonal Budget and MFP guidelines, PLM could suggest styles, price points, volume, colourways, specs, sourcing options, costs, timelines, margin, and ecological impact for each product and collection, by way of Generative Design by Impact? It won’t generate the definitive collection but, via comparison, enables rapid decision-making to narrow options. It allows the creative team to focus on the most innovative, profitable, and sustainable line based on all variables in the current situation.
However, the opportunity and constraint to delivering these powerful tools are the same for all AI and ML models – deep and accurate data. And this brings us full circle to the need for streamlined processes, science-based measurement of primary data, and carefully designed integration, to ensure generative AI models can access the comprehensive and accurate data required to provide a precise output.
It’s a fascinating time, and I can’t wait to see the evolution of Generative AI across the fashion supply chain. Whether fashion companies are focussing on merely complying with sustainability legislation or taking competitive advantage of Generative AI, plenty of ‘old-fashioned’ manual work is required to digitise the fashion supply chain. Once this foundation of connected data across the supply chain has been created, those businesses looking for competitive advantage can choose from hundreds of potential new use cases for Generative AI.
The fashion industry is running out of time, but it’s never too late to start. By starting now, brands, retailers, and their supply chain partners can get ahead of legislation, avoid related penalties, and turn a problem into an opportunity with the help of generative AI.
After originally training and working as an engineer, Chris joined a fashion services and technology company 30 years ago to implement ISO9001. Since then, he has helped over a hundred fashion brands, retailers, sourcing agents, and manufacturers to optimize their processes, supported by innovative technologies and concepts, working in offices, showrooms, and factories worldwide.
.jpg)
.jpg)
