Fashion Sustainability in the Era of Artificial Intelligence
Fashion may not be as glamorous as it sounds. We have been long romanticizing the industry with supermodels, fabulous designer looks, and luxury accessories. Few people paid attention to its environmental impact.
Cited from Bill Gate’s new book, “How to avoid a climate disaster”, 51 billion is how many tons of greenhouse gases the world typically adds to the atmosphere every year. A whopping 2.1 billion tons of those greenhouse gases come from the fashion industry, which’s 4% of the global total, which is more than double of the emission percentage attributed to transporting passengers and freights by airplanes!
If we must find a way to achieve carbon neutrality by 2050, the fashion industry needs an overhaul revolution. Many Brands and entrepreneurs are coming up with innovative ways to address this issue. As the second-largest global fashion retailer, H&M implemented conscious collections in 2012, where products are made out of more than 50% sustainable or recycled fabrics and labeled with green tags. Other brands also showed their commitment by offering discounts to their customers when they recycle their cloth. This circular fashion concept is well demonstrated by the success of the online secondhand fashion retailer, thredUP. Other companies like Queen of Raw, a New York-based company, is offering a second life for deadstock fabrics and textiles. My friend Megan’s company Clovo Brand is making bio-degradable pantyhose for women.
But, are these enough? Are these good enough to reach zero emission? If you are familiar with the fashion ecosystem, your answer will absolutely be NO. The fashion supply chain is a quite complex system. To make a cotton T-shirt, we start with growing cotton, harvesting the cotton flowers, making yarns and fabric. Then the fabric goes through dying, pattern cutting, and sewing. Eventually, it will be packaged, shipped, and displayed at the retail store close to your house. The ecosystem consists of manufactures, brands, wholesalers, retailers, and consumers. Where does artificial intelligence fit in? There are many existing marketing companies that are using AI to build customer profiles and recommendation algorithms that help brands and retailers run the targeted advertisements. Unfortunately, they don’t help with reducing the industry’s environmental impact.
Then, how can we solve the fashion sustainability problem with artificial intelligence? There are three folds to this question. The first part is to understand the status quo of the fashion industry. The second part is to know the capabilities and limitations of artificial intelligence. The last part is coming up with creative solutions to solve the problems.
First, let’s take a closer look at fashion’s waste problem. According to U.S. Environmental Protection Agency’s estimation, the generation of clothing and footwear waste was 13 million tons in 2018, which was 4.4% of total Municipal Solid Waste. And 70% of those waste ended up in landfills. Plastic microfibers shed from synthetic clothing into the water supply account for 85% of the human-made material found along ocean shores, threatening marine wildlife and ending up in our food supply. Nearly 70 million barrels of oil are used each year to make the world’s polyester fiber, which is now the most commonly used fiber in our clothing. But it takes more than 200 years to decompose. All these shocking facts deserve our thoughts and actions.
Next, what are the capabilities and limitations of AI? Many people have the misunderstanding that AI is this magical thing that will make all their problems go away. I’m here to tell you that AI can be really silly unless you use it right and feed it with large and quality data. Any AI model operates on the basis of a reward function. The goal of the AI model is to maximize the reward and minimize the penalty with some constraints. AI is really good at automating low compassion and routine work, such as driving, speech recognition, and transactional messaging. All the tasks have one clear reward function. The AI’s sole purpose is to fulfill that one goal and find the most optimal way to do so. AI is not competent to do high compassion and creative work, such as story-telling and brand-building. For the tasks that require low compassion and creativity or high compassion and routine, the best results are seen when humans and AI join forces.
Lastly, I see a number of promising use cases of AI for fashion sustainability. In the dying process, AI can be used to predict color depending on humidity, chemical composition, and fabric data. This can reduce the waste result from defective or inconsistent color. In the pattern cutting process, AI can be used to optimize the cutting strategy so that minimum fabric waste is generated from pattern cutting. In the fashion buying and merchandising process, AI can help fashion buyers and planners predict the demand for products in the next season and derive insights from omnichannel data. Fashion buyers can make informed data-driven buying decisions. This will largely reduce the overproduction and overstock waste. In the e-commerce shopping process, AI can be used to predict the size that’s most compatible with the customer’s figure. This can create value for brands with reduced return rates and shipping costs.
I am Kai Cui, founder of carbontag.co, the first carbon emission calculator for consumer fashion products. Find me on Linkedin at https://www.linkedin.com/in/cui-kai/
