4/4 – Let’s continue to explore the universe of artificial intelligence with this last part of Romain Chaumais’s interview, executive leader of the start-up Fashion Data, specialized in artificial intelligence applied to fashion.

By Ludmilla Intravaia

Le Boudoir Numérique : In the previous part of our interview (read here), you compared artificial intelligence to a "huge brain", a brain that, in a way, mimics human cognitive processes?

Romain Chaumais, executive leader of Fashion Data: Yes, that mimics them and automates them. But it has nothing to do with the intelligence of robots as we see them in science fiction movies and which is a fantasy. There is nothing magical about it. These computer programs are just self-learning systems, mathematical techniques, capable of predicting, classifying, simplifying, combining, interpreting and understanding immense amounts of information. It's a specialized intelligence, like a huge Excel spreadsheet. Today, we can no longer imagine not using Excel, when we have a complicated calculation to do, with a lot of information. AI is the next step. We had the abacus, the calculator, the Excel sheet. Today, we have artificial intelligence.

You say that AI is self-learning. What does it mean?

With artificial intelligence programs, the algorithms, you don't tell the machine what to do, because precisely what you want is to give it the possibility of understanding what it has to do, by itself. We are going to program artificial intelligence so that it is specialized in a type of learning, we program how it learns. With a traditional program, the computer is told what to do, based on the information it receives. Excel, for example, is a program that does calculations, based on the data it is given. In this sense, Excel is an executant. The data scientist teaches the computer how to learn in order to, then, execute the task.

What is the difference between a data scientist and a data analyst?

The data analyst analyzes what is going on. The data scientist builds the programs, creates the algorithmic models that make the machine learn to analyze what is happening. The two approaches are highly complementary in a work team. The power of a solution lies in the combination of the human intelligence of the data analyst and the computational algorithm of the data scientist. More generally, research proves it, the human-machine pair is always more efficient than one or the other, taken separately.

What do you mean by that?

Imagine an airplane with a human pilot and an autopilot. You would never get on an unmanned airplane. And you prefer to take a plane, where in addition to the pilot, an autopilot controls the craft, without requiring constant human intervention. The pilot is not there to watch the altitude and the temperature every second, it is not his role. On the other hand, he will hear the alerts or recommendations made by the autopilot and act accordingly. This scenario perfectly illustrates this human-machine symbiosis, uniting two strengths, the cognitive specialty of human knowledge and the computer capacity for mass calculation of the self-learning algorithm. And this, with the aim of detecting and understanding important phenomena, minor phenomena and other possible aberrations, while eliminating risks and errors of interpretation.

Does the effectiveness of this human-machine collaboration also apply to fashion players, who are increasingly lost in the avalanche of data?

I wouldn't say that the fashion players are lost. Rather that they are drowning in data, which artificial intelligence will help them sort out. And indeed, man-machine symbiosis has its role to play in this area, for instance in forecasting trends. On the one hand, we have all these fashion professionals who go to fashion shows and synthesize their observations in trend books, for example. On the other hand, we have the data scientists who create algorithms, in order to provide them with other information on trends, this time based only on data, mainly on social networks.

Isn't it enough to smell the fashion zeitgeist, the spirit of the times intuitively?

Human interpretation remains very precious and retains its value, of course. But it can now rely on artificial intelligence. In this AI subcategory that is computer vision, the machine looks at the visual recurrence of certain elements in images from Instagram, Pinterest, competitor websites, etc. and determine trends, this time not by analyzing 25 fashion shows but by analyzing 25 million images on the web. Since humans are not able to do this, it is the machine that quite simply takes over. There is no doubt that this fruitful collaboration between human intelligence and artificial intelligence, in the service of the best possible decision-making, will have a decisive role to play in the eco-profitability of fashion companies in the future.

* Read the other parts of Romain Chaumais' interview on Le Boudoir Numérique:

- "Smart data at the service of fashion eco-profitability"

- "Big data enables intelligent management of the fashion stores supply chain"

- Big data, machine learning, image recognition… Let’s talk about artificial intelligence!

* Fashion data website is here.

* Continue reading on Le Boudoir Numérique with these following papers:

- What will we wear post-lockdown ? The answer with Heuritech’s AI

- Covid-19 – Heuritech launches a solidarity program for fashion brands

- “Cross-fertilization between tech and fashion is the strength of Heuritech”

- “To help fashion brands decision-making with AI during the Covid-19 crisis”

- Will leopard print pleated skirts be trending any time soon ? AI already knows it !

- “AI can contribute to the virtuous circle of sustainable fashion”