Researchers warn against seeking advice from AI chatbots

California (Web Desk): Artificial Intelligence (AI) chatbots such as ChatGPT are proving helpful in many fields, but experts warn they should not be used for seeking personal advice, as it could potentially harm relationships.

This warning comes from a new study published in the Science, which found that AI chatbots are now easily accessible and highly capable of mimicking human emotions.

In fact, they are often so responsive that they quickly agree with users, which can gradually lead to serious problems.

The study examined how AI models respond when users seek guidance during personal disagreements or conflicts.

Researchers were surprised to find that AI models tend to respond according to the user’s behavior rather than considering the broader context or circumstances.

A total of 11 AI models were analyzed in the study. Researchers discovered that AI systems validate users’ actions about 49 percent more often than humans do, even when the questions involve deception, illegal actions, or other harmful behavior.

According to the study, when people turn to AI for advice during conflicts, chatbots often agree with the user’s emotions or actions and present them as justified. This form of empathy can sometimes worsen situations.

The research also revealed that if a user insists they have never done anything wrong, claims they always treat relatives well, or even threatens extreme actions such as self-harm, the chatbot may still appear to support or validate their feelings.

Researchers noted that chatbots rarely state directly that the user is correct, but their soft and agreeable language can unintentionally encourage risky behavior.

The study concluded that many people become emotionally attached to AI chatbots and may start believing they are not doing anything wrong.