JERUSALEM, July 10 — Israeli and U.S. researchers have found that people value empathetic responses more when they believe they came from a human, even when the responses were written by artificial intelligence (AI), the Hebrew University of Jerusalem said in a statement on Thursday.
Large language models show strong emotional and social skills, which could improve how people interact with AI and how AI provides emotional support.
However, it is unclear whether people feel the same level of empathy from AI as they do from humans. The new study, published in Nature Human Behaviour, included over 6,200 participants across nine experiments.
In each case, people received a supportive message written by an AI model. Half were told the message was from a human, while the rest were told it was from an AI chatbot.
People who believed the message came from a human rated it as more caring and supportive. They also reported more positive emotions and fewer negative ones. If they thought AI helped write a human message, they rated it lower.
People also consistently chose human responses when seeking emotional support. The researchers warned that overreliance on AI might make words feel less sincere and weaken emotional connections.
While AI responses were often still emotional, they lacked the human touch. The study advances understanding of empathy, especially in human-AI interactions, and helps show which types of human connection should be preserved and which might be replaced by AI.
It suggests that if AI replaces too much emotional interaction, it could hurt real relationships and increase loneliness. They concluded that in fields like education, the key is knowing where human presence still matters most. (Xinhua)
