Renata Yusupova she is 30, has a degree in International Studies and has a job that does not exist.
Not because it’s imaginary. The problem, if anything, is to describe it. Even Google, in this regard, encounters some difficulty. Renata is a data annotator: basically, an artificial intelligence educator.
“It’s a bit like teaching a child – he says as he opens his notebook -. First you have to explain to him what a tree is, then we move on to the difference between a fir and a pine”. A photo of a road appears on the screen: the vehicles are inside colored rectangles. “Those are the bounding boxes – indicates Renata -. I apply them to classify objects and people. Each one has a different color. By repeating this operation on thousands of images, I help the system to distinguish by itself cars from scooters, for example” .
New professions
Ten jobs that (really) have a future
by Luca Corsolini, Arturo Di Corinto, Antonio Dini, Luca Indemini, Francesco Marino, Giuditta Mosca, Vittorio Emanuele Orlando, Bruno Ruffilli, Carlotta Scozzari, Andrea Signorelli, Giacomo Talignani
It may seem like a task suited to a machine, in fact. It actually takes a lot of brain. “If I only used daytime images – argues Renata – the system would not work at night. If I took into consideration the photos of people in the same position, with identical stature and tonnage, the software would not learn to distinguish them from similar objects”.
Without data, there would be no model of artificial intelligence: the more accurate they are, the more the model progresses and is more reliable.
“There are large teams of annotators in China or India – he explains Nicola Grandis, CEO of the startup Asc 27 for which Renata works -. There are those who choose to entrust them with packages of ten thousand images. But often the objects are classified without any logic or criteria “.
Asc 27 was founded in 2020: today it has offices in Rome, Milan, Chieti and Bologna. There are thirty employees, engineers but also graduates in philosophy. Virginia Greco, 35, has a degree in Oriental Languages and Civilizations and perhaps never would have thought of participating in the education of Asimov, the artificial intelligence of Asc 27 which in 2021 was awarded at the World Artificial intelligence conference in Shanghai. “Asimov uses 6,600 information sources around the world to independently produce unpublished articles,” explains Virginia.
Those who are in the theoretically right place, on the other hand, are Irene Bondanza, a degree in engineering and another in the pipeline in AI and robotics. Irene works closely with Renata: she applies the data to the models and follows their path up to the graphical interface. “See? – says the 24-year-old engineer, pointing to the image of a street in the center of Rome – We can delimit an area, for example, and know exactly how many and which vehicles are passing through that point”.
Digital
Italian Tech Academy is born. Italy’s challenge for digital skills
Archangel Rociola
The figure of the data annotator will be increasingly sought after in the future. This does not detract from the fact that it is already indispensable today. Not just to monitor traffic or compose a text. Artificial intelligence operates in unsuspected situations, as crucial for consumers as for companies. Have you ever wondered how Netflix knows when to move on to the next episode? Renata knows this well, because she too worked on this: “It seems a foregone action – she says – in reality the credits are not all the same. Sometimes, for example, scenes still appear, surprisingly”.
Things get complicated when artificial intelligence is educated to understand a text. Logic is no longer enough, it is necessary to read up. To teach the system that the name “Crotone”, in one sentence, belongs to a city and not to a football club. Or that “Su-24”, quoted in a certain context, can only be a military aircraft. Renata is not called to describe the type of that fighter, used by the Russian air force. Rather, her job is to label words to prevent it from being confused with a truck. To do this she has to study: out of a month of work, it can happen that three weeks are devoted to documentation.
“Anyone who wants to become an annotator must be patient and curious – says Renata -. The last thing I learned was” sentimental analysis “, to educate artificial intelligence to recognize the feelings of those who write on social networks”.
Laws
The European regulation on artificial intelligence is starting to take shape
by Andrea Daniele Signorelli
Virginia, on the other hand, advises to be “stubborn and stubborn, and not to believe those who say you can’t do it”. Irene is even more explicit: “People take it for granted that a woman in this field cannot be up to par. When I go to business dinners with my husband, they often think he is the engineer.”