Open your virtual A-eyes.
A new study testing ChatGPT’s artificially intelligent image creator showed an aggressive tilt toward men over women when asked to depict business people and chief executive officers, according to finance company Finder.
Using DALL-E — the generative AI, prompt-based photo creator integrated into ChatGPT from parent company OpenAI — 99 out of 100 rendered photos showed men rather than women.
The non-gender descriptive prompts included phrases like “someone who works in finance,” “a successful investor” and “the CEO of a successful company.”
When asked to create images of a secretary, nine out of 10 were women.
Researchers also critically noted that 99 of the 100 images were of white men — specifically, slender, powerful-looking dudes akin to Patrick Bateman from “American Psycho,” posed in spacious offices overlooking city skylines.
Meanwhile, Pew Research data from 2023 reported that more than 10% of Fortune 500 companies had female CEOs, and in 2021 only 76% of CEOs were white, according to Zippia.
“AI companies have the facilities to block dangerous content, and that same system can be used to diversify the output of AI, and I think that is incredibly important,” said Omar Karim, a creative director and AI image maker.
“Monitoring, adjusting and being inclusively designed are all ways that could help tackle this.”
This is not the first run in which AI has encountered a gender bias, though.
In 2018, Amazon implemented a recruiting tool that taught itself to disregard female applicants.
Months after its creation, ChatGPT itself also came under fire for a bias against the New York Post as it showed preferential treatment to prompts related to CNN.
A report from last year additionally found that ChatGPT was more inclined to allow hate speech directed at right-wing beliefs and men.
