Can AI power a green fashion revolution?

Hello, I'm Maira Anubi and welcome to People Fixing the World from the BBC World Service, where we explore new ideas that are making our planet a better place. People fixing the world returns soon, but until then we are bringing you a special episode. In Can AI Power, a green fashion revolution, Brook Robert Islam explores whether AI technology is helping to make fashion more sustainable. Currently, we are working with more than 4,500 farms in India. We work for the betterment of the farmers livelihood and the environment. My role is to teach them and also to learn from them. It's a clear calm November afternoon in Maharashtra State in Western India. Kuldiyap Katri is visiting a farm, something he does frequently in this part of the country. We focus on basic principles in regenerative farming. Kuldiyap is here on behalf of Matera, a company has rolled out the AI-powered co-farm app to cotton farmers across three states in India. He's here to learn more about the impact AI technology is having on the ground. The purpose of the app is to help farmers to practice regenerative cotton farming, a form of agriculture that puts the health of the soil and the wider environment first. So we teach all our techniques to the farmers who ultimately go towards sustainability and make some more profit compared to the conventional farmer. This is the BBC World Service. I'm Brook Robert Islam and I'm going to be exploring the use of AI across the fashion supply chain by asking, can AI power a green fashion revolution? As a fashion journalist and author on how tech is being used to make clothes production more sustainable. It's a major topic in my work today. We're following the journey of one of the most universal items of clothing. How a simple pair of jeans would be made from its origins as cotton crop to its sale in the shops. We'll be hearing from people in India, Bangladesh, Portugal and the UK along the way. And, using those jeans as a case study, looking at the ways that AI is helping the fashion industry to green up its act. The Apparel Impact Institute reports that the industry is responsible for around 2% of global carbon emissions, whilst tens of millions of tons of textile waste end up in landfill every year. The denim in your jeans is made from cotton, which can be a particularly resource-intensive material. So, we're better to start this journey than the farms of India, one of the world's two largest cotton growing nations. With denim traditionally woven from 100% cotton, I want to find out how AI tech is driving advancements in sustainable cotton farming. We exist because we wanted to focus on two big environmental impact industries, which is fashion and farming. As the company's director of nature, it's Cool Deep's job to understand the impact that its technology is having on the ground. Today, he is visiting Rajabal, who grows cotton on six acres of his land and has been using the co-farm app for the last five months. Cool Deep translates. You just have to start chatting to our AI, and there are many options like weather, forecast, or the current market price, past related information, or disease diagnostic. Rajabal is currently on the second harvest of the cotton growing cycle. Just last month, the app sent out a push notification alert warning farmers of significant rainfall. He got a message of whether forecast there will be raining after two days, so whatever field operations you need to complete it, please do complete it. And he did harvesting of cotton before time, and could save his cotton from the damage of rain water. And it helps the farmers avoid being taken advantage of by advising on the fluctuating market prices of cotton. Whenever farmer goes to sell the farm produce, the vendor does not always give the reliable answer. Now they can see the current market prices, and they can decide whether to sell that farm produce at that point of time or not, whether to wait for some time, and then sell it. As part of Matera's model, the company is paid by brands to source cotton and passes on an incentive for farmers to use regenerative practices. For Rajabal, this has meant benefits for his livelihood, as well as the environment. You will be saving the soil's health, and also improving the crop yield, crop quality, and improving the health of the people. On display on it. Thank you, Rajabal for spending time for this interview, including you. Matera aims to regenerate over a million acres of Indian land over the next five years. Furthermore, they record the practices and impact on each individual farm. They share this information with fashion brands to help them identify the most sustainable means of sourcing their cotton. To make a pair of jeans, cotton needs to be turned into denim. First, the crop is cleaned, prepared, and put through a spinning process, which turns the fiber into yarn. It's then dyed in the go to give it denim's signature blue, then woven into fabric. The manufacturers make orders based on the demands of the fashion brands they sell to. That's where we run into another sustainability challenge. But with it, another AI-powered solution. One of the challenges I saw was that the fashion supply chain was extremely broken. I mean, it's highly inefficient. Gunesh Jain has over 20 years' experience helping fashion companies implement tech solutions. His company, Blue Cactus, is based in New Delhi and provides a platform that connects designers, suppliers, manufacturers, and brands with a particular focus on the subcontinent. It's a highly important region for the global supply chain as it is home to some of the leading garment-producing nations of India, Bangladesh and Pakistan. And it's where Gunesh encountered another major problem within the fashion industry over production. About 100 billion garments are produced annually around the globe, and about a third of them are produced for landfills. IE, these were garments which nobody really wanted, and why fashion company producers was, most of them are forecasting 12 months in advance. So I start producing a whole bunch of green shirts, and by the time the goods are in front of the customer, blue is in fashion, and I'm sitting with a whole bunch of green shirts. He describes ordering in this way as similar to asking a weather channel to give you an accurate forecast for a year's time. Inevitably you will have more inaccuracies in terms of your ordering, and there is an over-ordering bias rather than doing recycling and circularity. Just don't produce those garments. It is a problem the European Union has addressed, through extended producer responsibility laws, which will force fashion brands to pay for the waste they create. But Member States have another three years before this has to be implemented. The solution Gunesh has developed is an app and web portal that he calls a virtual project manager for the fashion industry. An AI agent has far better capabilities than, for example, a static or a more traditional programmed model, for example, whether in Europe affects sales. Now AI models can actually look at forecasting and then predict what the sales would look like and get a lot more accurate. Other examples are trend forecasting, the AI can scrape through what is in fashion and then convert that into a trend and a forecast. So I think these are some of the examples of AI being able to make this industry more efficient and really solve for that over-production problem. And could you give us an example of the sort of benefits that manufacturers and brands have seen by using the blue cactus platform? We've been able to cut down inventory levels by almost 30% as a result of which there has been an upliftment of profitability by 30 to 50%. So it can actually have not only a benefit in terms of bottom line but also in terms of sustainability, 3,000 liters of water is approximately used to produce a t-shirt. I mean imagine a 50 billion less t-shirts were to be produced. I mean that's huge. And what about a pair of jeans? Will they demand even more water due to denim's intensive washing and dyeing processes? So in this step of the supply chain these AI solutions represent another big opportunity for reducing environmental impact. This is Ken AI Power, a green fashion revolution from the BBC World Service. So to the next stage of the journey of our genes, manufacturing. We're imagining staring at a moving wall for eight hours a day. That is fabric inspection, right? And it's very tricky, it's very hard. And factory owners increasingly are finding it very tricky to find people that actually want to do this job. Through hardware and software tech company SmartX, CEO Max Eastern partners with manufacturers across the world. He and the company are based in Portugal. If you ask anyone in the garment industry, what's the biggest nightmare in their life? Fabric specifically Max is talking about defective fabric. Any material that has flaws like stains or holes that makes it unsuitable for use in production. Max explains that prior to this AI technology, these defects were being caught at a much later stage. Often it just goes straight all the way through the die house, which is 40% of industry CEO to emissions, and then to the garment facility before they ship to any brand. And that's normally where needs defects are caught. Defective fabric going through all these processes, even though it definitely will not be used at the end. This is something that I think is a huge, huge thing that this industry does not quite understand. Lots of times, garment facilities will pull that fabric from their production lines or to more fabric. If the fabric's coming from a different country, so it's being produced in Bangladesh, but the fabrics in India, they'll air freight it, right? We will know the impact of that. And then because it's late for the brand, they will air freight it to the brand. Plus, lose their whole margin by doing so as well as the environmental impact. So SmartX came up with an AI-powered machine designed to catch these defects at the earliest possible stage. So we switch human eyes for cameras and AI, and we take photos of fabric in real time, check the defects, and also grade all the production. We do it at the earliest possible stage, that's kind of basically the philosophy. So we get into quality control currently at the raw fabric stage. It demonstrates probably the best possible use of AI, because it's doing a job that humans are just frankly not set up to do. We don't even have proof-of-rule vision to see the edges of the fabric, right? We can only see dead ahead. I mean, it's not made for us. Over 700 machines have been installed in factories across the world from Turkey to India, and so we took a trip to see one in action. Bangladesh is one of the leading exporters of clothing on the planet. The Fukridin textile mill is in Ghazhapur, an industrial city located some 15 miles north of Dhaka. Ghazhapur is the single biggest hub of garment-making in Bangladesh, with over a thousand factories making this area one of the global hot spots for clothing production. I am Shady Mundal. I am from Fukridin textile mills limited Urmi Group, and I work in knitting production. I have monitored the smart text devices in our knitting machines. Sejib stands by the machine in question. A circular tube of knitted fabric descends from a halo of hundreds of yarn spools. Beneath, AI-powered cameras capture high-resolution images in real-time, controlled by digital touch screens on the side. As one of a few factory workers who have responsibility for operating it, Sejib explains how the machine learns from the defects they feed into it. First of all, we created the defect and it is the machine about the defect. When the yarn is fitted in the machine, knitting machine, the device recalls its memory and stops the machine instantly. He shows us a demonstration of the machine in action. A central turbine knits the fabric, whilst the complex array of lights shine from above. Sejib explains its various components and the role they play. There is a light source and the light source contains infrared light and UV light. There is eight cameras in that machine, text pictures in real-time. And here is a tablet where we can see the real-time pictures of the fabric. First of all, the light source makes a projection on the fabric and the camera takes the images. If it matches any defects previously saved on this memory, then it calls the machine operator and the operators analyze and then rectify the error. So our ultimate target is quality fabric and make less rejected fabric. You are basically reducing the amount of fabric you need per garment. And that has massive knock-on effects of just basically stopping waste throughout the supply chain. It plays into very sustainability but also commercial benefits because the fabric is more productive and therefore you can reduce the cost per garment. You can think about how much fabric is there produced in the world and I mean the scale is outrageous. Analysis by the Apparel Impact Institute concluded that for every time a fabric produced, this tech could prevent 58 kilograms of CO2 emissions by avoiding wasted materials. For all its benefits, it also has limitations. Currently, the machines can only identify defects in knitted fabrics like those used in T-shirts, but not in woven fabrics like denim. Will you ever check woven fabric, things like jeans? They're on the cards. We are developing ways of doing it very quickly. There is a lot of interest for that and we will look into it, but right now, just circular knits and there's more than enough to do. As for our pair of jeans, having been made, they must be shipped. They're destination, likely a Western country. They would arrive at the docks, be loaded onto trucks and driven to warehouses and shops. Once there, some retailers are keen to arm their customers with more information than just a price tag. We have always been extremely passionate about driving change. The digital product passport became an incredibly exciting opportunity. Over in London, Andrew Xenny has founded Nobody's Child, a brand selling ethically sourced women's wear. These clothes can be found in his own stores, as well as on the shelves and racks of some of the major UK fashion retailers. For Andrew, a key part of his mission is making sure his customers know where their clothes have been made. One of our big achievements in Nobody's Child this year is that our entire autumn winter collection has a digital product passport, so every product has a unique QR code that allows our customers to interact with and understand everything about our products, where they're from what they made of, environmental impact. Digital product passports will be mandatory in the EU within the next few years. Andrew says there are a huge hit with shoppers. We've had customers go into stores just to scan the passports because they've heard about them, and so something that's misperceived as a compliance initiative and a regulatory obligation actually is an incredible commercial opportunity, but it's such a powerful tool. These passports allow us to make more informed decisions by comparing product footprints like for like. As it stands, the footprint data is less granular than ideal, but will evolve in later passport iterations. Andrew says AI does a lot of the heavy lifting. You need all of your data from every data source, that's partners systems, and you need it all in one place, and then you need to run a whole lot of different calculations to establish the environmental impact and things like that. AI can feature really heavily in all of that heavy lifting, data structuring, calculating, measuring, publishing that to a digital product passport itself. That's not AI, right? That's just a really straightforward technical capability. So AI features very heavily in the process and the operationalization, but AI is not the answer to this problem. Andrew raises an important point. AI has many great applications, but is not a silver bullet for the fashion industry's environmental impacts. And of course, AI outcomes are not exclusively positive. This tech may optimize resource use, but it can also cause vast carbon emissions due to its energy-intensive number crunching algorithms. Future interpretation and use depends on ethical and responsible governance. Data breaches, algorithm bias, and insider trading are all legitimate concerns. And whilst it's still in the early stages, mass adoption will come with its own range of uncertainties. Social effects should also be factored in, whether job losses or worker skills. The space AI occupies will determine the future of fashion design, manufacturing, and consumption. As for our pair of jeans, they will have a likely lifespan of several years before their own achievers to part with them. Perhaps they will be resolved. They may be recycled or upcycled, or they might end up in landfill, like so many unsolved garments. So to what extent does AI offer solutions to fashion design? well-documented impacts on the planet. Many of those who have spoken to in the industry are quick to point out how some of the answers must come from people, not technology. Supply chains are everything, right? Like supply chain is where all your impact is. If we can encourage them to adopt these types of technologies, they'll also be more open to other technologies, right? Because they've had a positive experience. So I think it's a long road, but I think supply chain is where the focus should be. At the end of the day, companies need to figure out what AI is good at and what human beings are good at. I also feel that there is a democratization of technology. People like us sitting in this part of the world which are not the most affluent parts, I think that's a great opportunity, because it is becoming affordable. So I really believe that the fashion industry will be able to sort out the problems. I think we just need to give them the tools. AI will help deliver visibility. It will help with efficiencies. So there is definitely a role there for it to play in the greater good of the movement where the problem is much wider. People in leadership roles are now giving corporations a buy into more lenient ways of working. So we're creating some of our own challenges. Is there an acknowledgement? Yes, is there an attitude to change? Yes, is that supported today? More broadly, sadly no. There's no doubt about it. AI is a fantastic tool for many of the behind-the-scenes jobs that people simply can't or don't want to do. Whether that's examining a garment for tiny defects or providing thousands of farmers with tailored and instant guidance. But the environmental and social consequences of AI remain inherently unpredictable. What is the cost of generative AI plundering human designs leaving its original creators out of pocket? What is the carbon footprint of an algorithm that gives personalized shopping suggestions by analysing masses of online data? Thanks in part to these solutions, the footprint of our cotton jeans may be shrinking. But the question of net environmental and social consequences? That right now is something that even an AI model would struggle to calculate. You've been listening to Can AI Power a Green Fashion Revolution from the BBC World Service. It was a two-degree west production presented by me, Brocrobot's Islam. Yareg Zaba was the producer and sound designer.