Google Image Search Results are ‘Amplifying Gender Stereotypes’
A new research study has found that online images diplayed on search engines and social media may be reinforcing gender stereotypes.
Previous academic works have tended to focus on text but the new paper published today by Berkeley Haas claims that gender bias is over four times stronger in images than text.
“Most of the previous research about bias on the internet has been focused on text, but we now have Google Images, TikTok, YouTube, Instagram — all kinds of content based on modalities besides text,” says Solène Delecourt, a professor and author of the paper. “Our research suggests that the extent of bias online is much more widespread than previously shown.”
Previous research has indicated that the human brain better retains information from images than it does from text. In one of the experiments conducted, study participants who looked at gender-biased images — as opposed to those reading gender-biased text — demonstrated significantly stronger biases even three days later.
Methodology
The researchers employed a “novel” technique which involved taking the top 100 Google images for various search times like “doctor” or “neighbor” and then compared it to online texts from Google News by employing large language models.
The researchers then asked 450 participants to use Google to find descriptions of occupations relating to science, technology, and art. One group used Google News to find text descriptions while the other group used Google Images.
After selecting either text or images, the participants rated which gender they most associated with the given occupations.
They then had to quickly sort various words into gender categories — a test they underwent again after three days.
The subjects who had worked with images displayed much stronger gender stereotypes than the people who used text — even three days after the fact.
“This isn’t only about the frequency of gender bias online,” says Douglas Guilbeault, a professor and author. “Part of the story here is that there’s something very sticky, very potent about images’ representation of people that text just doesn’t have.”
And when looking at which genders do which jobs, the researchers found that disparities were far more pronounced on Google Images than the official numbers kept by the U.S. Bureau of Labor Statistics.
Visual content online also feeds AI image generators which will pick up on similar biases.
The paper can be read here.
Image credits: Header photo licensed via Depositphotos.
