As we engage more frequently with generative artificial intelligence, it becomes important not only to grasp the fundamentals of how it works, but also to recognize that it exposes us to certain cognitive biases that can shape how we perceive and use it. Below are five key phenomena to keep in mind during your interactions with AI.

 1. Algorithmic bias

Although generative AI relies on statistical models and may appear neutral, its outputs are not automatically objective. The algorithms are trained on human-generated data, which can contain or amplify existing biases (sexist, racist, etc.). If certain groups are underrepresented in the data, the results may reinforce distorted or discriminatory assumptions, leading users to trust conclusions that seem logical but may in fact be misleading or harmful.

2. The black box effect

Generative AI can deliver coherent and complex outputs, yet the steps between a user’s request and the system’s response remain opaque. Whether a user is a beginner or an expert, this lack of transparency — the “black box effect” — can create confusion or uncertainty and influence how much trust they place in the system’s results.


3. Subjective interpretation of results

A user’s background — including their knowledge, past experiences, and expectations — heavily shapes how they interpret an AI-generated answer. Two people may read very different meanings into the same output.


4. Overestimating or underestimating AI

Users may fall into the trap of overvaluing AI’s abilities, assuming it can solve problems flawlessly and without human oversight. This is known as automation bias. The opposite can also occur: underestimating AI’s real capabilities, which may lead to missed opportunities or unnecessary skepticism.

5. Social acceptability of AI

Generative AI is a powerful technological shift that raises unresolved ethical questions related to security, autonomy, privacy, and democracy. How society views and accepts this technology influences how individuals approach and interact with it, shaping their comfort level and expectations.