that AI




that react,


Dr. Ahmed Zaidi
Co-Founder and Chief Executive of Hyran Technologies,
Visiting AI researcher at Cambridge University

Artificial intelligence is more and more frequently being used by brands for trend forecasting, taking into consideration variables such as consumer behavior, runway trends, and various trends in fabrication, colors, and other themes. What many do not realize, however, is that AI trend forecasting is contributing more to our textiles waste crisis than many other waste-generating culprits, like post-production disposal by consumers. I am deeply familiar with this issue, as I come from a family of textile manufacturers that have directly felt the consequences of inaccurate forecasts. 

AI forecasting is a noisy guess masquerading as an objective analysis — and manufacturers know it. My family suffered from not only an excess waste problem but was often put in financial distress when the forecast against which we bought and planned our raw materials was inevitably wrong. 

Attempting to predict the future with AI technologies is futile — particularly for production cycles exceeding more than a few seasons. Human behavior is largely unpredictable, and a virtually infinite number of unforeseen variables may impact the success of products once they reach stores, contributing to excess inventory and thus waste. That’s before we even account for everything that happened before those products were available for sale.

While predictive AI forecasting is a massively imperfect system, AI systems still can vastly improve sustainability efforts for brands by instead mitigating inventory risk. This can be achieved by narrowing production lead times, and allowing cross-collaboration between brands and their suppliers by unlocking agility and flexibility in the supply chain. To drive a sustainable and efficient supply requires multi-dimensional optimization, something humans struggle with but AI systems thrive at.





Macro- and micro-trend forecasting have faced scrutiny by sustainability advocates who argue ever-changing seasonal textile trends and fast fashion are contributing to our waste crisis. But few have argued that AI modeling for predicting consumer behavior months or even years into the future is at fault, and even fewer have provided a clear alternative to this industry-wide practice. 

We have the tools to use AI for multi-dimensional supply chain optimization to shorten production lead times instead of using it for predicting future demand, which is a huge contributor to our tens of millions of tons of manufacturer-generated waste each year. Overproducing generated by AI forecasting — a kind of crystal ball system for attempting to predict unpredictable consumer behavior — is not only hurting our environment with textile waste, but it’s also detrimental to manufacturers and their working conditions. Inaccurate forecasts force manufacturers to make last-minute changes and work longer hours to meet the changes demand.

Regulators in regions of the US and abroad have proposed that brands pay up for their contributions to our global waste crisis, including by funding textile recycling programs by paying for the volume of products they produce. These policies, or “extended producer responsibility” (EPR) regulations, could look similar to programs used to reduce waste for things like batteries and mattresses — among other materials that can be difficult to recycle. They attempt to mitigate overproduction by holding the brands of these goods responsible. But as noted by Bloomberg, and as is often the case with these types of fees, it’s likely these companies would attempt to offset such fees by passing the costs to consumers.

That’s another way that AI could work to overhaul the production process and ultimately curb waste in ways that regulatory fees may not: this kind of modeling could help eliminate production waste before products reach consumers. By harnessing AI to eliminate waste and excess costs to brands from the earliest stages of product development, brands could potentially avoid damaging price-gouging to consumers while also minimizing the fees they pay for any waste they generate under such policy frameworks. 

Each year, around a trillion dollars is lost to markdown programs, making the economic benefits to brands using AI in this new, more proactive way clear. Furthermore, using AI for multi-dimensional supply chain optimization could also have far-reaching benefits to the fashion industry at large, including contributing to greater sustainability efforts and redistributing those costs to improving workers’ wages. For instance, if enough small- or medium-scale brands buy into this collaborative AI-driven information sharing at the earliest stages of product development, they stand to benefit tremendously from their combined spend at the materials level — perhaps even meeting or exceeding the spend of a singular large-label brand.


versus reactive


Convincing brands to break tradition with industry-wide practices is a clear hurdle, particularly for established heavyweights for whom change to existing systems is a slow process. Fashion houses with long-established manufacturer relationships may be more reluctant to explore alternative methods to supply chain navigation. But again, I’d argue there’s tremendous potential to smaller and medium-level brands as well as lower margin brands who, through collaboration and transparency made possible with AI optimization modeling — rather than futile forecasting — could see significant gains and recaptured profit by eliminating potential markdowns or post-production waste.

By viewing production from a risk perspective, it’s possible to approximate a brand’s carbon footprint not only from a single-product lifespan, but rather multiple lifespans: circular, remaking and secondary, and re-owning. As our textile waste crisis comes under increased scrutiny, brands have new opportunities to harness AI to ethically produce, both to their own benefit as well as to consumers and the environment.

Do you agree with this?
Do you disagree or have a completely different perspective?
We’d love to know