Our thinking

Internet’s gender bias

24 October, 2025

We often think of the internet as a mirror, showing us the world as it really is. But a recent study from UC Berkeley, Stanford and Oxford suggests it’s not quite that simple. 

Researchers analysed 1.4 million images and videos from sites like IMDb, Google Images, Wikipedia and more. They found a consistent pattern: women were systematically portrayed as younger than men, whether in images of celebrities or of everyday professionals. For celebrities, women’s photos averaged an age of 25, men’s 44. 

But it didn’t stop with pictures. The study also explored the language used online. Words like “expert”, “leader”, or “senior” were far more likely to be associated with older men. Women were often framed differently, as less experienced, less authoritative. 

These patterns might seem subtle, but they matter. Because the internet doesn’t just influence what we see, it also trains the algorithms that increasingly shape our decisions. The same study found that AI tools like ChatGPT gave higher scores to older men when asked to rank CVs. Younger women fared better but were still less likely to be “promoted” by the system. 

That’s not a glitch, it’s a feedback loop. One that starts with biased images and words, and ends with real-world outcomes, in hiring, in media, in opportunity. Because the way we represent people, online and off, doesn’t just reflect our world. It helps shape the systems that run it. 

By Justine Bahoumina