Artificial Intelligence Suffers from Human Prejudice!

Amazon tried an AI-powered recruitment programme but found that the software favoured men. A Microsoft chatbot transformed into a sex-obsessed, neo-Nazi machine in just 24 hours. The error rate for facial recognition reaches up to 35 per cent for women of darker skin tones, while for white men it is less than one per cent. An experiment from Swinburne University revealed that people express negative feelings significantly more often towards women in high positions than towards men. In call centres, a female "soothing" voice is used more frequently (just as IBM prefers it in debate competitions). It turns out that people prefer to be spoken to by female voices.

Charles Cuvelliez, who lectures at the Free University of Brussels, wrote about these and other incidents in the newspaper Les Échos.

Scientists agree that these traits are transferred from the data on which the networks are trained.

The article concludes with the following reflection. According to the World Economic Forum, it will take until the next century to achieve true gender equality. And it is possible that in the case of artificial intelligence, we may have to wait even longer for gender equality.
What do you think?

Source: https://www.lidovky.cz/…/caste-zavady-lidske-predsudky-budo…

Source: https://www.ceskenoviny.cz/…/les-echos-do-umele-int…/1722367

Original article: https://www.lesechos.fr/…/cercle-190391-opinion-pas-de-dive…

European "Focus on Human Beings": http://blog.alphai.cz/svetovy-zavod-v-robotice-nezaostane-…/
Artificial Intelligence and Ethics: http://blog.alphai.cz/umela-inteligence-a-etika/

A set of seven ethical principles that Google commits to uphold in the development of artificial intelligence: http://blog.alphai.cz/soubor-sedmi-etickych-zasad-ktere-se…/

Artificial intelligence surprised in a debate competition: http://blog.alphai.cz/clovek-zatim-zvitezil-umela-intelige…/