A new Artificial Intelligence (AI) algorithm, developed by scientists at the University of Hong Kong, can only create enormously realistic photos from sketch drawings. Researchers say the new algorithm can be used to identify suspects in police investigations.
Despite the concerns of names like Tesla and SpaceX CEO Elon Musk, Artificial Intelligence (AI) technologies continue to penetrate every aspect of our lives. AI, which has a place in many sectors including security, health, military, automotive and transportation, is becoming more and more powerful.
According to the information reported by the website of New Scientist, Hongbo Fu and colleagues from the University of Hong Kong developed an algorithm that can instantly convert a very simple sketch drawing representing the face of a person into portrait photography.
The researchers think that the artificial intelligence can be used in police investigations to quickly analyze images of suspects or to model characters in games and movies.
Fu and his team used more than 17 thousand portrait photos shared publicly on the Internet to train the AI algorithm. Over time, a doodle (simple drawings that can consist of random and abstract lines) or an algorithm that has learned to turn the sketches into a real face has achieved tremendous accuracy.
However, this artificial intelligence algorithm, which is focused on face identification, has some stereotypes, i.e. prejudices, originating from the data sets used during the learning process. Because almost all the faces created by the algorithm are from the white race.
“Most of the portraits created point to white-skinned people. We do not have any control over this,” says the name behind the study, Hongbo Fu.
Meanwhile, IBM CEO Arvind Krishna, who wrote a letter to the US Congress last week, said the company will no longer offer general-purpose facial recognition technology or analytics software. Expressing that they are against all face recognition technologies offered for mass surveillance, racial profiling, basic human rights violation or any purpose that is not consistent with company values, Krishna said that they will not conduct R&D in the field in question.