‘Fashion is passion’, it is said in English; Fashion is passion and that, at least in principle, seriously limits the possibilities of Artificial Intelligence (AI) as a tool for companies in the sector; Machines do not know about feelings.
But the bottle can be seen half full or half empty: fashion will always need the human factor – be it fears about the future of employment in this business – and, at the same time, machines are a perfect complement to that passion and imagination that puts The human being when designing, in this case garments.
That was the main thesis defended by Angel Castan, director of IBM’s Cognitive Retail during his talk at the Fashion Tech conference held recently in Madrid. “Technology is not creative, so it has to rely on the human being. Machine and man must work together, “he said. And as proof he explained in detail two experiments he has already carried out with the fashion industry Watson, the division of Artificial Intelligence of IBM.
One of them was – literally – during the Metropolitan Museum of Fashion’s latest fashion show, the Met Gala, one of the most important annual couture quotations. There the model Karolina appeared with what IBM called the ‘cognitive suit’, a garment resulting from the joint work of the computer giant with Marchesa, American fashion firm specializing in wedding dresses.
The experiment started with the idea of reflecting in an evening dress five feelings: joy, emotion, courage, passion and curiosity. To determine which colors best express those feelings, Marchesa’s designers resorted to IBM’s AI, previously ‘fed’ with a huge database of the brand’s earlier designs, to ensure that Watson’s recommendations were consistent with Marchesa trajectory, and data on the effect of colors on the mood.
The next step, once the colors were analyzed, was to decide the fabric of the garment. IBM technology made a first selection of 40,000 types of fabrics, which was reduced to 150 and finally to 35, taking very into account the premises of the designers. The final choice among those 35 options was Marchesa.
There was the final touch, the most striking. The dress was fitted with 150 LED lights – as if they were flowers – that were illuminated, of different colors, depending on the comments that the garment was subject to on Twitter..
Another use of Watson for fashion that IBM is already experiencing is, beyond design, trend prediction. In this case, the pilot was done in collaboration with fashion designer and textile entrepreneur Jason Grech.
Watson worked in this case as a research tool with a speed and power unattainable for the human being. He analyzed 10 years of images in fashion magazines and social networks like Instagram, Pinterest and Twitter. Taking into account the likes of each image and how they shared. Even understanding conversations in the networks about it.
The aim was to determine fashion trends regarding colors; In particular, the reactions to 1,027 tones (in more than 700,000 images) were analyzed to finally determine which ones were more likely to become fashionable. “Without technology, it is impossible to do this work,” said Castán, who is not trying to replace the designers, but to increase the tools of analysis within their reach.
But IBM’s goal is not to make eye-catching action with the world of haute couture. There are applications of Artificial Intelligence much more practical for the day to day, as its use in the virtual assistants in the e-commerce. If you sell fashion, the virtual assistant has to know fashion, just like the shop assistant.
By analyzing images, Watson can know which colors are sold the most, what are the most common clothing combinations and, in general, the style that the buyer is looking for, to suggest a fuller cart. And, as he boasted in a small Castan demo, he even takes into account weather forecasts, “something IBM has known a lot since buying The Weather Channel” last year. So if you tell the machine that you are going to a party and want to release something, it is able to make suggestions based on the weather.
The improvement of these virtual assistants can increase the percentage of online fashion sales, which in the second quarter of 2016 (latest available data) was, according to the National Commission of Markets and Competition, a relatively modest 5 – 6% (by business volume). However, the key to the sector is not to sell more on the Internet, but in general. Hence the obsession with the so-called Omnichannel. The term became known several times during the day Fashion Tech.
On the other hand, Lorente, product manager on The Spot Services, a company that among other services digitizes physical stores to increase sales and improve customer awareness, warned that many times businesses confuse being multi-channel (being on And offline) with an authentic Omnichannel. “The shopping experience is the advantage of physical versus online commerce, but it has to be noted that experience, that advantage,” he said. As long as they know how to adapt to digitization and take advantage of their strengths, physical clothing stores will never go out of style.