Many experts in this field claim that artificial intelligence systems are full of biases and racist biases… This has been revealed by research conducted by three universities.

You will also be interested in this

(on video) 10 most dangerous artificial intelligence threats Artificial intelligence has made possible advances in health, research and…

By force of circumstances, an artificial intelligence (AI) is trained on millions of documents, and its “intelligence” is constructed from these sources. However, these data relate primarily to white American culture. According to various experts, this would result in uncontrolled racist bias.

The research was conducted by the US universities of Washington and Carnegie Mellon, but also by Xi’an Jiaotong in China. The results of his research were published in a memoir. In each case, it was noted that generic AI tools displayed particular social and political biases in relation to the location where the data were collected.

racist prejudice

examples ofFirst abundantly so, according to ProPublicaAn AI tool used by US courts to assess the likelihood of crimecrime Criminologists predicted twice as many false positives for black defendants (45%) compared to white defendants (23%).

Recently, Vision A.I. GoogleGoogle It is estimated that people with darker skin thermometerthermometer Portable carried a rifle. When he was light-headed, it seemed as if he was recognizing an “electronic device”.

An AI in China could decide to put you in jail

a journalist from buzzfeed — which eventually retracted its article — says it used the generative AI MidJourney to create distinctive Barbies from 194 different countries. Result of operation:

  • German Barbie wore SS uniform;
  • South Sudanese Barbie had a gun firefire ,
  • Lebanese Barbie standing in front of a ruined building.

English CNIL concerns

in the United Kingdom,Information Commissioner’s Office (ICO) believes that Artificial Intelligence ” harmful consequences on people’s lives » And so companies in the sector should take measures to mitigate these risks.

The problem is much bigger than it appears at first glance. Increasingly, artificial intelligence will be integrated into systems capable of influencing certain decisions, especially the hiring of individuals. Therefore, it would be regrettable if some were excluded from the beginning on the basis of arbitrary criteria.

Categorized in:

Tagged in: