Can AI be sexist? Amazon showed that it is possible

Artificial Intelligence (AI) is being used to improve efficiency in different companies and facilitate the performance of some tasks for human beings. However, this technology is not infallible and Amazon proved it when it was forced to reject an AI that was sexist.

An algorithm that was being tested as a recruiting tool by the company proved to have a preference for one of the sexes.

This system was trained with data submitted by a large number of applicants over a period of 10 years, which were mostly men. This resulted in the AI ​​considered men more able than women to obtain a job in the company.

The system's developers, in an interview for Reuters, explained that the algorithm had the purpose of reviewing candidates' job applications and giving them a rating of 1 to 5 stars. "Literally they wanted me to be an engine where I'll give them 100 resumes, spit out the first five and hire them," said one of the IA designers.

However, it is clear that it still needs improvement, since the IA understood that male candidates were preferable to women. Amazon has not yet responded to the claims caused by the defective technology.

The system has been developing for years. By 2015, it was clear that it did not qualify the candidates in a neutral manner. In fact, AI began to penalize curricula that included the word "women." The program was edited to make it neutral and launched this year, but it continued to present sexist tendencies.

This increases the doubts about how reliable Artificial Intelligence can be for some tasks. According to a survey conducted by the software firm CareerBuilder, about 55% of human resource managers in the United States say that AI will play an important role in the recruitment of workers over the next five years. problems of biases are solved as soon as possible.

This is not the only case of partialized AI. In May of last year, a program generated with Artificial Intelligence used by a US court, had prejudices against people of color, as it marked them as twice as likely to reoffend in criminal activities as white people.