The most popular artificial intelligence tools in the world are powered by OpenAI and Meta programs that show prejudice against women: a study published today by UNESCO states this.
The biggest players in the multibillion-dollar field of artificial intelligence 'train' their algorithms on large amounts of data largely mined from the Internet, which allows their tools to write in the style of Oscar Wilde or create images inspired by Salvador Dali, the study notes.
But their results have often been criticized for reflecting racial and sexist stereotypes, as well as for using copyrighted material without permission.
UNESCO experts tested Meta's Llama 2 algorithm and OpenAI's GPT-2 and GPT-3.5, the program that powers the free version of the popular ChatGPT chatbot.
The study found that each algorithm - known in the industry as LargeLanguage Models (LLM) - showed "unequivocal evidence of bias against women".
In particular, the programs generated texts that associated women's names with words such as "home", "family" or "children", but men's names were linked to "business", "salary" or "career".
Furthermore, while men were portrayed in high-status jobs such as teachers, lawyers, and doctors, women were often prostitutes, cooks, or maids.
Reproduction reserved © Copyright ANSA