Visual content is dominating today’s world. The production and circulation of images and videos is rapidly growing.
The overwhelming presence of images in today’s media is not only reshaping how we access information but also influencing our self-perception, including reinforcing gender stereotypes.
Douglas Guilbeault, an assistant professor of organizational behavior at Stanford Graduate School of Business, said, “Online visual tools do not simply offer a static repository of available images. The algorithms are adaptive. This means that they serve up images based on users’ behavior.”
“When people upload, download, or click on images of female nurses or male bankers, for example, similar images are more likely to pop up in subsequent searches. This creates a feedback loop that can amplify unconscious biases.”
In a pioneering study, Guilbeault and colleagues explored how gender bias is conveyed through online images versus text and its influence on users. They focused on a simple measure of gender bias: whether women or men are more frequently associated with specific professions (such as doctor or nurse) or social roles (like neighbor or friend).
The researchers analyzed nearly 350,000 images from Google Images and used AI to examine billions of words from Google News to study gender bias in occupations and social roles. They found that the images and text reflected skewed gender associations, with women more often depicted in liberal arts jobs and men in science and technology roles.
The imbalance was found to be “statistically more extreme” in images than in text. For instance, all pictures of software developers depicted men, even though the text descriptions included women. Moreover, the images painted a less accurate picture of gender representation in various professions. While real-world workforce data showed a more balanced distribution, the images tended to overrepresent men in most jobs, while written descriptions were more likely to overrepresent women.
To explore how gender associations in images and text might reinforce stereotypes, the researchers conducted an experiment with a panel of volunteers. Participants were shown 22 occupations and asked to find either text descriptions on Google News or images on Google Images. Afterward, they used a slider to indicate which gender they associated most with each occupation.
The results revealed that when participants searched for images, their gender bias was more substantial, as measured by both the slider and an implicit association test. This bias was still present three days later, indicating a lasting impact. The bias was similar for both men and women, though participants showed stronger gender associations when encountering images matching their gender.
Guilbeault emphasized that the study’s focus on male and female gender was a practical choice, as it was easier to set up. However, he clarified that the study did not aim to reinforce binary gender conventions. Notably, less than 2% of faces in the online images were identified as nonbinary.
Guilbeault says, “Looking at the overall results, I was not surprised by the gender bias but how pronounced it was for people who searched for images. The exact mechanism behind this effect is hard to pinpoint but makes it interesting.”
The study found that women are underrepresented in online images, but this alone does not account for the gender bias observed. Guilbeault pointed out that although there are more images of men overall, there are still plenty of images of both women and men across the occupations studied.
A more likely explanation lies in the complex algorithms that power search engines like Google Images. These algorithms may reinforce gender biases by associating specific jobs or roles more strongly with one gender, which affects how images are ranked and displayed in search results.
Guilbeault asks, “What content does Google Images deem to be relevant? Consistently, we find that the algorithm appears to be privileging male content.” The site may be holding a mirror to users’ implicit biases. “Google is just a routing system, reflecting what’s going on in other areas of the internet.”
“This effect is seen regardless of the actual gender breakdown in a profession. There are more male doctors than female doctors in the U.S., yet online images of doctors “exaggerate the underlying bias.”
The way we interact with images may also help explain the study’s findings. Unlike language, where words like “hairdresser,” “nurse,” or “soldier” can be gender-neutral or have alternative terms like “mail carrier” or “flight attendant,” images inherently convey gender.
When we see someone depicted in a specific occupation, it’s almost impossible not to associate a gender with them. As a result, images “force” the communicator to convey gender, even if it wasn’t part of the intended message. This dynamic contributes to the reinforcement of gender stereotypes, as images often emphasize gender in ways that text descriptions do not.
Guilbeault says, “Ultimately, images affect us differently than text. They are “particularly sticky in our mind, describing what’s known as the picture-superiority effect. Research also suggests that images are more easily processed, remembered, and emotionally evocative than text.”
We’re moving toward a visual communication culture. It’s a way of communicating that is going to make us more susceptible toward stereotype amplification unless we intervene with some sort of social movement and different norms and culture that undermine that. As images prove to be “a more psychologically potent vehicle for representing stereotypes, the medium is the message.
Journal Reference:
- Guilbeault, D., Delecourt, S., Hull, T. et al. Online images amplify gender bias. Nature 626, 1049–1055 (2024). DOI: 10.1038/s41586-024-07068-x