Artificial intelligence (AI) has the potential to radically improve our lives, from increasing global food production to helping to treat disease and shielding us from cyber attacks.

But here’s the rub: there are serious concerns that advances in technology will leave millions of people without jobs. A recent report by consultancy firm PwC found that in the UK, 30% of all jobs are at risk, including 2.25 million in the wholesale and retailing sectors.

For those in the fashion industry, this concern may be exacerbated by the development of an AI model capable of learning how a person likes to dress and inventing new outfits based on their preferences.

Image: Flaticon

The system was developed by scientists at the University of California, San Diego, and Adobe Research.

It works by combining two AI algorithms. The first is a convolutional neural network (CNN) that uses Amazon purchase data to learn a person’s preferences for certain products and suggest similar items – a feature that is already used by many online shopping sites.

However, the CNN data is then combined with a general adversarial network (GAN), which can be used to generate new images of items that are deemed most consistent with a person’s preferred style.

In other words, it has the potential to choose a variety of outfits based on products you have either searched for or purchased in the past. This is in addition to helping design new clothes and modifying existing outfits, the researchers say.

Left are suggestions for similar items based on Amazon purchase data; and right are synthetic images of items that a consumer might like based on their preferences.
Image: University of California San Diego

While the tool is clearly exciting, it is not quite at a stage where it can actually choose a new outfit or replace a shopping assistant – but the idea has piqued the interest of several major technology companies.

Fashion forward

Last year, for example, Amazon’s Lab126 research and development arm developed a similar algorithm designed to analyze images and use the information to generate new styles.

Speaking to university magazine MIT Technology Review about this AI, Cornell University professor Kavita Bala said: “There’s been a whole move from companies like Amazon trying to understand how fashion develops in the world. This is completely changing the industry.”

In a similar vein, Google recently created an AI model that can accurately predict which images a human would like according to their aesthetic appeal.

According to Google, its Neural Image Assessment tool can be used to compare the quality of different versions of the same image, which may have been distorted in various ways, while also being used to enhance the perceptual quality of an image.

It could also be used to improve picture-taking in real-time and assist in the editing process, its developers claim.