How AI is driving a 'distortion of reality' when it comes to how women are presented online

Women successful enactment are consistently portrayed arsenic younger and little experienced than men online and by artificial intelligence, a sweeping caller survey has found. 

Researchers studied 1.4 cardinal online images and videos, arsenic good arsenic 9 ample connection models trained connected billions of words.

They discovered that women are systematically presented arsenic younger than men successful thousands of occupation roles and societal positions.

"This benignant of age-related sex bias has been seen successful different studies of circumstantial industries, and anecdotally, specified arsenic successful reports of women who are referred to arsenic girls," said University of California, Berkeley's Solene Delecourt, who co-authored the study.

"But nary 1 has antecedently been capable to analyse this astatine specified scale."

The quality was astir stark erstwhile women were represented successful high-powered positions similar main executives, astronauts oregon doctors, contempt determination being nary important property imbalance successful reality, according to US census data.

"Online images amusement the other of reality. And adjacent though the net is wrong, erstwhile it tells america this 'fact' astir the world, we commencement believing it to beryllium true," said Douglas Guilbeault from Stanford, who co-authored the study.

"It brings america deeper into bias and error."

Read much exertion news:
Bitcoin's terms astatine grounds highs. Is it sustainable?
Actor's girl begs radical to halt sending AI videos of him
Almost 15 cardinal teens usage vapes globally - report

The researchers besides utilized ChatGPT to make astir 40,000 CVs and recovered the AI assumed women were younger and little experienced portion standing older antheral applicants arsenic much qualified.

In different experiment, they recovered that participants who viewed women successful occupation-related images estimated the mean property for that occupation to beryllium importantly lower, portion those who saw a antheral performing the aforesaid occupation assumed the mean property was importantly higher.

For jobs seen arsenic female-dominated, participants recommended younger perfect hiring ages, portion for male-dominated occupations, they recommended older hiring ages.

"Our survey shows that age-related sex bias is simply a culture-wide, statistical distortion of reality, pervading online media done images, hunt engines, videos, text, and generative AI," said Ms Delecourt.

OpenAI told Sky News it has conducted extended probe into systemic bias successful ChatGPT and recovered that little than 1% of the AI's responses bespeak harmful stereotypes.