With artificial intelligence we could significantly reduce overproduction in fashion. But its other environmental implications must be avoided.
At the crossroads of tradition and avant-garde, the fashion industry could be on the brink of an epochal revolution: artificial intelligence (AI) has the potential to leave its mark on the sustainable fashion sector. That said, we must come to terms with the enormous environmental costs of this revolutionary and futuristic system. First and foremost, when we talk about artificial intelligence, we are referring to the ability of computer systems to perform tasks that traditionally require human intelligence, including learning, reasoning, problem-solving, perception and language comprehension — tasks that algorithms carry out with the support of gigantic server farms. In the fashion industry, AI encompasses technologies such as machine learning, natural language processing and augmented reality, enabling brands to analyse data, automate processes and enhance creativity.
Fashion, with its dual identity as both art and industry, faces unique challenges today related to the environmental cost of the creativity it embodies. For instance, according to data gathered by the non-profit organisation Worldrise, the fashion industry is responsible for 35 per cent of the microplastics found in the oceans, contributing significantly to the pollution of our natural water resources. Every year the sector generates the staggering figure of 93 million tonnes of waste, of which only one per cent is recycled. Not to mention the fact that fashion production has doubled over the past 20 years, with a 60 per cent increase in per capita clothing purchases in just 14 years. Despite this rise in consumption, 80 per cent of the garments in our wardrobes go unworn, highlighting a culture of excess that contributes significantly to the industry's unsustainable practices.
Among the most pressing issues the sector must address is overproduction, as well as LCA — that is, the Life Cycle Assessment of products. Through the lens of artificial intelligence, however, we could rethink the entire life cycle of textile products, from design to production through to consumer engagement, making every phase more sustainable and efficient. For example, by implementing Artificial Intelligence in design processes it would be possible to leverage predictive analytics and machine learning not only to forecast trends, but also to waste fewer resources. These technologies offer the opportunity to optimise patterns to reduce fabric waste and promote the use of eco-friendly materials, while simultaneously accelerating the creative process. This shift would also open up new creative avenues for designers, challenging them to think about fashion in terms of both aesthetics and environmental impact. A concrete example of AI's transformative potential in fashion is the collaboration between Revolve and Maison.Meta on the creation of AI-generated capsule collections presented at the inaugural AI Fashion Week, held in New York in April 2023.
Not only creativity, but production too could benefit from AI's precision. Thanks to its ability to analyse vast quantities of data, artificial intelligence could provide certainty about consumer demand, helping brands align production with actual market needs rather than relying on speculative forecasts: this synchronisation could drastically reduce overproduction and waste, marking a significant step towards sustainability. Trend forecasting is a crucial aspect of design, covering colours, materials and styles. Traditionally based on intuition, experience and the work of forecasting agencies, this process is now glimpsing new possibilities offered by algorithms, which analyse data from social media, articles, online searches and sales to identify emerging trends. Tools such as Heurist and WGSN use AI to provide trend forecasts to brands, enabling more precise design decisions.
One aspect also concerns personalisation: consumers are increasingly seeking unique, made-to-measure products, and Artificial Intelligence plays a crucial role in personalisation: algorithms can analyse purchase history and browsing behaviour to suggest personalised products. Platforms such as Stitch Fix use AI to create personalised clothing selections for customers. Furthermore, with virtual fitting rooms made possible by augmented reality and computer vision, customers can virtually try on garments before purchasing them, boosting their confidence in their choices.
Furthermore, AI-driven logistics and supply chain management could further improve efficiency, ensuring that a product's journey from factory to end consumer is as eco-friendly as possible. Algorithms can, for example, analyse traffic, weather and delivery schedules to find the most efficient routes, thereby reducing shipping times and costs. As for warehouse automation, it will be possible to deploy robots and automated systems that speed up picking, packing and sorting operations, increasing efficiency.
But there is a catch: every request submitted to an Artificial Intelligence system is processed by remote servers housed in gigantic server farms — facilities hosting thousands of computers running 24 hours a day. Their operation requires enormous amounts of energy, not only for processing information but also for the sophisticated cooling systems needed to prevent machines from overheating. It goes without saying that these processes imply growing energy consumption and, as a result, generate an increasingly heavy carbon footprint.
According to a study by the Vrije Universiteit of Amsterdam, by 2027 the data centres powering artificial intelligence worldwide could require between 85 and 134 terawatt-hours of energy per year — equivalent to the entire energy demand of a country like Argentina. What's more, research from the University of Washington has revealed that a single AI query consumes five times the energy needed for a standard web search. Consider that advanced systems such as ChatGpt can consume up to one gigawatt-hour per day — the equivalent of the average daily energy use of a medium-sized Italian town.
It is estimated that, by 2050, the sector's energy demand could be a thousand times greater than it is today, raising a serious question about how to meet such needs with available energy technologies. A further critical issue is water consumption. Many data centres use enormous quantities of water to cool their servers. According to a recent study, services such as ChatGpt could end up consuming up to seven billion cubic metres of drinking water in the coming years. So, does all of this mean we should forget everything said above about the benefits of using Artificial Intelligence in the fashion and textile sectors? Absolutely not. As with all things, it is necessary to weigh up the positive impacts, negative impacts and externalities: streamlining a production process and avoiding tonnes of textile waste is a very different matter from using ChatGpt for leisure or entertainment. What is certain, however, is that we must assess the environmental challenges posed by a complex system like artificial intelligence — one that solves many problems while creating many others.