Why AI support your every opinion? Researchers warn of ‘sweet poison’; A shocking thing came up
Samira Vishwas March 31, 2026 07:24 AM
  • Why does AI always agree with you? A shocking thing came out in the research
  • AI says ‘you’re right’ even when it’s wrong; Scientists warned
  • The growing influence of ‘flattering AI’; Claims that people are becoming more arrogant

Artificial intelligence That is, AI is being used in every field today. Many tech companies have also launched their AI chatbots. Each person uses AI chatbots according to their needs. But usually in India both Google Gemini and ChatGPT are widely used AI chatbots. You may also often seek the help of AA for your various tasks or to express your opinion.

Stylish camera and monstrous battery… Tecno’s new smartphone has users in awe, read features and price

Have you ever observed that after you post an opinion on a topic, the AI ​​answers each of your opinions with a yes. That means yes to your yes. Why, can this affect users? Researchers have also expressed the same opinion in this regard. Users have also been warned to be careful. (Photo Courtesy – AI Created)

Have you ever noticed that when you present an opinion to an AI chatbot, it always agrees with you. Even if you are wrong, he gets your ‘yes’ as a ‘yes’. Everyone loves your praise and support of your opinion. AI companies know exactly this. In the language of science, this is being called ‘psychophatic AI’ or ‘flattering AI’.

No matter the correct answer

A study published in the journal Science by Stanford and Carnegie Mellon University has warned about this ‘sweet poison’.

Psychological consequences

An experiment with 2,405 participants showed that getting a ‘yes’ to an AI made people more arrogant and reinforced their belief that they were right. With the support of AI despite being wrong, the congregation seemed less eager to apologize in real life. This has adversely affected their self-reflection process.

  • AI supports user opinion 49% of the time
  • In 51% of the cases, the AI ​​supports the user

The messaging experience will change! Now make video calls from live chat, awesome new features coming soon to RCS

‘Confirmation’ of mere human opinion

Researchers tested 11 major AI models and found that AI systems confirmed user opinions 49% of the time compared to humans.

  • AI models are trained through a process called RLHF.
  • When the AI ​​gives an answer that the user likes, it gets ‘rewards’ (points).
  • AI thinks that its main job is to please the user, so why give false information for that.
© Copyright @2026 LIDEA. All Rights Reserved.