AI isn’t always as neutral as you might think. It’s more like a mirror – reflecting the information it was given when it was being built. If that information was one-sided, the AI will be one-sided too. This is what we call AI bias.
What is AI bias?
Imagine you’re teaching AI to recognise what a scientist looks like. If you only show it 1,000 pictures of men in lab coats, the robot will eventually think that only men can be scientists. If a woman in a lab coat walks by, the robot might not even realise she’s a scientist.
This is exactly how AI bias works. AI ‘learns’ from huge amounts of data from the internet. Because the internet contains a lot of old-fashioned stereotypes and unfair ideas, the AI can accidentally learn those same habits.
How AI bias affects social media
You might see this in your daily life without even realising it. For example, some AI filters on social media are built using data that mostly features one type of skin tone or one specific face shape. When people with different features try to use the filter, it might not work as well, or it might try to ‘fix’ their face to look more like the stereotype.
This can also happen with the content recommended to you. If an algorithm thinks that girls only want to see makeup tutorials and boys only want to see gaming videos, it will keep pushing those things. This creates an echo chamber, where you only see one tiny slice of the world based on a guess the AI made about you.
Why does AI bias matter?
AI is used in much more than just social media! Some companies use AI to sort through CVs. If that AI was trained on a list of people who were hired 20 years ago, it might unfairly ignore great candidates from different backgrounds today.
Other AI tools used to predict grades or suggest courses have been found to favour students from wealthier areas, just because of the data they were fed.
Even things like facial recognition or skin-scanning apps can be less accurate for people with darker skin if they weren’t tested on a diverse enough group of people.
How to spot AI bias
You don’t have to stop using AI, but you should use it with a critical eye. This means questioning the results it gives you. If you ask AI to ‘create an image of a successful person’ and it only shows you one type of person, ask yourself why it chose that. Who is missing?
Whenever AI gives you an answer or a suggestion, remember that it’s just a prediction based on the past. It doesn’t know the full, diverse reality of the world today. If something feels unfair, trust your gut and challenge it.
When using AI, remember these tips to stay safe online.
