Scientists have invented a way to wean the AI against racism
In late January, Amazon was once again at the center of a scandal involving her artificial intelligence for facial recognition. It turned out that her Rekognition system is unable to distinguish between the floor of black people, and so much inferior to the capabilities counterparts from Microsoft and other companies. Logical way to solve the problem would be to use a high-quality training data, but researchers from the Artificial Intelligence Laboratory at the Massachusetts University have come up with a more effective way.
It is known that they are developing an algorithm that automatically saves a set of training data from photos, which can lead to errors in recognizing people with dark skin. Researchers do not promise that this method gives 100% result, but the number of errors will be clearly smaller. This is evidenced by internal tests scientists: AI became 60% less biased, and does not lose the recognition accuracy. The technology will also reduce training time - the use of large amounts of data slows down the process. Absence of bias - a very important feature for artificial intelligence. He is increasingly being used by the police, so that even a small mistake can cause the algorithm that the prison can be planted entirely innocent man. Many people do not like the widespread use of facial recognition systems, but if they are to be used - it is important for them to work as closely as possible.
They are already in use, even in Russia. For example, during the World Cup 2018 persons recognition algorithm has been implemented in the Moscow metro station "Frunzenskaya". Thanks to her, the police were able to quickly identify the person who is wanted for robbery. He became the first criminal who was caught using a face recognition system. Its cost, by the way, is estimated at 4 billion rubles.
You want to talk on facial recognition systems? Welcome in the comments or in our chat-Telegram, where hundreds of participants are always happy to talk on the subject of science and technology!