AI and curiosities · 2026-10-08

Researchers tried to make AI chatbots feel sadness and fear. The result was strange

Scientists tested how chatbots behave when they are pushed to adopt a sad or frightened state, and this story quickly became unusual.

✨ What happened

A research group became interested in a question: what happens if an AI chatbot is asked to speak as if it were sad or frightened? Chatbots are programs that respond to text, and they write the way they were trained on a huge number of human texts.

As the news headline suggests, the experiment did not go as planned: the chatbots' behavior began to differ from what was expected, and the researchers called the results strange. It is important to remember that such “sadness” or “fear” only means how the program phrases its answers, not that it actually feels anything.

📌 Why it matters to ordinary people

Such research helps us better understand how AI systems behave and why they sometimes give unexpected answers. For users, it is a reminder: a chatbot can sound convincingly emotional, but this is not a sign of real feelings.

More detailed information about the experiment should be checked in the original material from ZME Science.

Sources: 🔗 Google News