Artificial Intelligence and the challenges of the fashion industries
by Francesca Mazzi* and Laura Palazzetti
The fashion industry has experienced a considerable growth in the last decade. Nonetheless, such growth came with numerous challenges concerning the sustainability of the business mode, that affects the worldwide community. This article aims at analysing to what extent the use of Artificial Intelligence can help in reshaping the business towards social and environmental sustainability.
The phenomenon of fast fashion
The term “fast fashion” captures well the rapidity that concerns the fashion cycle nowadays, as a response to an increasingly “disposable” society. Indeed, fast fashion items are those that embodied trends that are meant to be temporary because they are replaced rapidly. Traditionally, fashion houses used to work on two main collections, spring/summer and autumn/winter. Nowadays, fast fashion stores change their windows and their exhibited pieces monthly. Such phenomenon targets a well-defined consumer: a consumer that is willing to change his wardrobe often, spending a relatively small amount of money.[1] Indeed, the fast fashion consumer does not expect a particularly high level of quality and intends to use the item probably for an amount of time that is less than a season, and it expects the clothing to be out of trend relatively soon. Hence, fast fashion, in order to function, needs a continuous replacement of clothing[2] and it consequently produces an enormous amount of unused or unsold pieces that are not desired by the consumers even after a short amount of time.[3]
Looking at the impact of this phenomenon in Europe, according to the European Parliament research service[4], “About 5% of household expenditure in the EU is spent on clothing and footwear, of which about 80 % is spent on clothes and 20 % on footwear. It has been estimated that in 2015 EU citizens bought 6.4 million tonnes of new clothing (12.66 kg per person). The European Environment Agency (EEA) estimates that, between 1996 and 2012, the amount of clothes bought per person in the EU increased by 40 %. At the same time, more than 30 % of clothes in Europeans’ wardrobes have not been used for at least a year. Once discarded, over half the garments are not recycled, but end up in mixed household waste and are subsequently sent to incinerators or landfill”. Such business models are based on a linear economy, that can be summed up as “take-make-dispose-waste”.
Environmental impact
The fast fashion model has a tremendous impact on the environment, which is a direct consequence of the “disposable” model described in the paragraph above, together with other factors such as emissions related to shipping and online shopping. Collecting some data available online, the following figures can provide an idea of the scope of such impact: “global textile production has more than doubled in the past 15 years. Additionally, the total greenhouse gas emissions from production, at 1.2 billion tonnes annually, are more than those of all international flights and maritime shipping combined.”[5] Furthermore, “the fashion industry (…) is responsible for 10% of global carbon dioxide emissions, 20% of the world’s industrial wastewater, and 25% of all insecticides used in the industry.”[6] “It’s single-handedly responsible for 10% of global CO2 emissions, a quarter of all insecticides used and a fifth of the world’s industrial waste water.”[7]“92 million tonnes of textile waste is estimated to be generated by today’s fashion industry on an annual base. 80% to 90% of the EU furniture waste in MSW is incinerated or sent to landfills, with ~10% recycled. EU28 consumption of €68 billion, and consumption of ~10.5 million tonnes of furniture per annum.”[8]
Social impact
The fast fashion industry exploits labour. Some data can help understanding the scale of this sector. More than 50 million people in Asian developing economies are employed in the garment sector, where for decades some countries have based the economic growth on millions of low-paid workers. Nowadays, “more than half of the world’s textile exports and about 70% of its ready-made apparel exports” come from developing Asian countries. Several noticeable issues can be highlited, for instance, the role played by women. According to the “Sumangali” scheme, women who work in looms are locked up between 3 and 5 years inside the factories, with exhausting work rates and starvation wages. Payroll are usually paid by the “employers” at the end of the contract and is often used as a dowry for the wedding. [9] Yet, on the other hand, it is noticeable how this fast growth played a pivotal role in their empowerment and independency. In fact, the garment sector is one of the most female-oriented industries, considering that in Bangladesh the 85% of workers are female (in Cambodia the percentage arises to 90%). Are the new technologies replacing them or integrating and facilitating their daily-jobs?
Even with all the human-rights-related issues these workers are facing, working for the western big fashion brands lifted entire families out of poverty and upgraded the life conditions of entire areas of these countries. Hence, it is clear why the biggest fashion brands are used to relocate to countries such as India, Pakistan or Bangladesh to reduce production costs and take advantage of labour laws that are much more elastic and permissive than those of industrialized countries. The relocation is exploited above all by the brands that offer low-cost garments and that target the Fast Fashion market. Producers had to adapt to having cheaper and faster production chains, often subjecting the by-products to unbridled competition that has favoured wage compression and child slavery.[10]
So far, the manifacturing jobs have been cheap enough for western companies, surely cheaper than the implementation of AI technologies but that is changing and the consequences could be devasteting for workers. It may only be a matter of time before fast-fashion brands exchange the clothes-making process. In fact, according to ILO[11] “the average bangladeshi garment workers earn 68 us dollar per month making them the lowest paid sectors in Asia . Because the cost of labour is so low, oftentimes, human workers are less expensive and more capable than machines but as machines become more sophisticated and easily available their threat becomes real”. It is a vicious circle, having “made […] extremely difficult for manufacturers to produce according to the ‘ethical’ standards prescribed by lead firms. In other words, pressures that originate within buyer-lead firms’ intensified circuit of capital are ‘solved’ in the manufacturers’ circuit of capital, yet, in ways that disconnect them from the ethics of the former and threaten the sustainability of the entire value chain.”[12]
The potential of AI in the fashion industry
Artificial intelligence (AI) is a broad term used to describe different approaches to create “intelligence” in machines, but it does not refer to a single concept.[13] In fact, the term “intelligence” itself can be linked to human rationality as well as the mere concept of imitation of the functioning of the human mind. To simplify, various approaches exist within the umbrella term AI, and each of them is aimed at replicating certain human characteristics in machines, by making them act or react in a way that is as human-like as possible.
Technologies within the concept of AI range from machine learning to neural networks, to name two of the most relevant. Machine learning combines statistics and computer science in order to enable machines to autonomously process and learn from data. The algorithms are trained and work based on data, that allows them to produce predictive models and to continuously refined themselves proportionately to added data. Deep learning is an approach to machine learning where artificial neurons are connected in networks to generate an output, based on weights associated with the inputs.[14]Indeed, artificial neural networks (ANN) use the brain as inspiration: they are built with the aim of replicating the way in which neurons (i.e. nerve cells) in the brain receive inputs, process the inputs, and then produce an output (i.e. the activation of a synapse).
The field of AI is experiencing a flourishing growth, with machines impacting every aspect of everyday life and every industry sector. The fashion industry is part of it, and various companies are starting to engage AI to face some of the challenges described above. Although numerous remarkable initiatives related to AI in the fashion industry are in place, the potential of AI seems to hold way more revolutionising power to reshape the entire business model. Nonetheless, as it will be described in the final paragraph, AI alone would not be enough, as regulations and reforms are another necessary step towards sustainability.
AI to reduce the environmental impact
AI can help reducing the environmental impact of the fashion industry in many ways. Currently, AI is already engaged by many companies to reduce waste, for example, by analysing data and predicting inventory levels, as well as quantity of products needed by physical shops. Moreover, AI is also used to suggest sizes in online shopping to avoid returns, which clearly leave an environmental footprint. Moreover, trend detection at the initial stage of design could reduce forecasting errors by 50% and overall inventory levels by 20-50%.[15] One key element to reshape the fashion business model would be to switch from a linear economy to a circular economy. As described above, the linear economy is characterised by a “take-make-dispose-waste” approach. A circular economy could be resumed in two types of approaches: one with technical materials that would be a “make-use-return” and one with biological materials that would be “make-consume-enrich”.[16] Through the analytical approach, AI can design circular products, components and materials. Moreover, AI can operate circular business models through the combination of historical and real-time data from producers and users, using analytic tools as well as predictive tools to reduce waste.[17] Finally, AI could optimise circular infrastructures, by improving production processes as well as identifying the optimal networks and systems to redesign entire supply chains.[18] Moreover, AI can help enhancing the functioning of the second-hand market, which ultimately part of a circular economy model, by creating or stimulating existent peer to peer shopping app for clothing rental and resale, such as DEPOP.[19]
AI for social responsibility
AI together with other technologies such as blockchain can substantially help improving transparency of the supply chain and accountability. Transparency is a key factor to move towards social (and environmental) sustainability of the fashion industry. Indeed, by tracing origins of the raw materials, labour of the garments, policies of suppliers working at various stages of garment production, companies can be incentivise to monitor work conditions as well as social impact of their production. Moreover, the creation of labels such as “conscious product”, similar to what has been developed by H&M, could promote awareness amongst users and producers. Nonetheless, the mere use of AI might not be sufficient. Indeed, AI represents also a threat to human work, as it is able to replace human workforce. In fact, in order to avoid job losses, AI shall be accompanied by code of conduct and renovation of companies’ priorities.
The example of TOMS is of great interest in this sense.[20] TOMS is a company that produces shoes and that decided to set as main goal the improvement of life, and not the profit per se. For each pair of shoes sold by TOMS, a pair of shoes is delivered for free to people living in underprivileged conditions. Moreover, TOMS donate 33,3% of their net profit, and started operating other activities according to their core value, such as providing water, prescription glasses and safe birth kits.
In conclusion, during the last decade many scholars questioned themselves about the implementation of a new techonology for a more sustainable garment industrial sector growth and the implementation of human rights in this process . Yet, how this challenge has to reshape according to what has been called as the “fourth industrial revolution”? The development of technology is a fact, whether we like or not and artificial intelligence and robotic machines are increasingly replacing humans’ skills and capacities. The use of AI technologies undoubtedly brings important benefits to consumers during the entire purchasing process. It affects the moment of choice, potentially influencing our decision process and nonetheless it speeds the making process and even the delivery of the ready-made products. In addition, the automation potentially eliminates the wastes caused by human errors, optimizing the entire chain. Its is clear that governments need to be more actively involved and should develop mandatory policies to require companies to perform due diligence checks across their entire supply chains.
*Francesca Mazzi, Early Stage Researcher – f.mazzi@qmul.ac.uk
[1] Bruce, Margaret; Daly, Lucy (2006) “Buyer behaviour for fast fashion”. Journal of Fashion Marketing and Management. 10 (3): 329–44. doi:10.1108/13612020610679303.
[2] According to Sajn, N., ‘Environmental impact of the textile and clothing industry’, (2019), ‘Fast fashion constantly offers new styles to buy, as the average number of collections released by European apparel companies per year has gone from two in 2000 to five in 2011, with, for instance, Zara offering 24 new clothing collections each year, and H&M between 12 and 16’, 2-3 (2019)
[3] Choi, T., Liu, N., Liu, S. et al. Fast fashion brand extensions: An empirical study of consumer preferences. J Brand Manag 17, 472–487 (2010). https://doi.org/10.1057/bm.2010.8
[4] EPRS, Sajn, N., ‘Environmental impact of the textile and clothing industry’, (2019) available at https://www.europarl.europa.eu/RegData/etudes/BRIE/2019/633143/EPRS_BRI(2019)633143_EN.pdf
[5] Gossellin, V., ‘How artificial intelligence can help fashion brands be more sustainable’ (2019) available at https://www.heuritech.com/blog/articles/fashion-retail/how-artificial-intelligence-can-help-fashion-brands-be-more-sustainable/
[6] Byers, G., ‘Artificial Intelligence is Restyling the Fashion Industry’ (2020). Avilable at https://towardsdatascience.com/artificial-intelligence-is-restyling-the-fashion-industry-c2ce29acae0d
[8] Rudradeb, M., ‘How AI and Circular Economy can save the Fashion industry‘ (2019) avilable at https://medium.com/towards-artificial-intelligence/how-ai-and-circular-economy-can-save-the-fashion-industry-2dcdbeb0da86
[9] Arexxini, ‘Fast fashion e lavoro sfruttato: il vero valore dei vestiti a basso costo’. Available at https://www.arrexini2.it/fast-fashion-e-lavoro-sfruttato/
[11] International Labour Organization “Wages and Working Hours in the Textiles, Clothing, Leather and Footwear Industries” available at https://www.ilo.org/wcmsp5/groups/public/—ed_dialogue/—sector/documents/publication/wcms_300463.pdf
[12] Nikolaus Hammer, Réka Plugor Work, “Disconnecting Labour? The Labour Process in the UK Fast Fashion Value Chain” (2019) Employment and Society 33(6) https://doi-org.ezproxy.library.qmul.ac.uk/10.1177/0950017019847942
[13] Russel, S. J,; Norwig, P., ‘Artificial Intelligence: A Modern Approach’ (1995, 2020)
[15] Gossellin, V., ‘How artificial intelligence can help fashion brands be more sustainable’ (2019) available at
https://www.heuritech.com/blog/articles/fashion-retail/how-artificial-intelligence-can-help-fashion-brands-be-more-sustainable/
[16] Rudradeb, M., ‘How AI and Circular Economy can save the Fashion industry‘ (2019) avilable at https://medium.com/towards-artificial-intelligence/how-ai-and-circular-economy-can-save-the-fashion-industry-2dcdbeb0da86
[19] Kerins, J.; Wallace, M., ‘Finding the right fit: why AI is the solution to the sustainable fashion crisis’ (2020). Available at https://www.ibm.com/blogs/think/uk-en/finding-the-right-fit-why-ai-is-the-solution-to-the-sustainable-fashion-crisis/