Can AI Manipulate You Without You Noticing?
Imagine opening a social media app and finding a video that perfectly matches your interests. Then another. And another. You may feel that you are choosing freely, but the content you see has already been selected for you.
The invisible influence
Many digital platforms use recommendation systems to select content based on information such as your previous activity and interests. These systems influence which information you encounter, although they do not automatically control what you think.
A 2024 review of research on recommender systems explains that these technologies can influence the information people see online, potentially affecting their beliefs, decisions and actions.
AI can also make communication more personal. A system that adapts its responses to your interests may make its suggestions feel especially relevant or convincing. This does not necessarily mean it is deliberately manipulating you. However, personalisation can make its influence harder to recognise.
When does influence become manipulation?
Not every recommendation is manipulation. A system suggesting a film you might enjoy can be useful. The concern grows when a system hides important information, exploits personal vulnerabilities or encourages decisions that serve someone else's interests without making this clear.
The important question is not simply whether AI influences us. Advertising, friends and news have always influenced human decisions. The question is whether we understand that influence and remain free to make our own choices.
How can you protect your judgement?
Before accepting an AI recommendation, pause and ask yourself three questions:
- Why am I being shown this?
- What information might be missing?
- Would I make the same decision if I had not seen this recommendation?
These questions cannot eliminate every form of influence, but they can help you make more deliberate decisions.
AI does not need to control your mind to influence your behaviour. Sometimes, changing what you see is enough to change what you consider possible.
The question is: how often do you stop to question what is influencing you?
Reference
Deldjoo, Y. et al. (2024). Fairness in recommender systems: research landscape and future directions. User Modeling and User-Adapted Interaction, 34, 59–108.