People prefer human sympathy over AI, even

Even if AI produces the exact same empathic response as humans, people don’t value it too much because they know it comes from machines.

A new study involving more than 6,000 participants in nine experiments shows that the source of empathy is as important as its content, having a profound impact on how we incorporate AI into healthcare, education, and daily emotional support.

This study, published in Natural Human Behavior, shows that people always rate AI-generated responses as disagree, supportive, and emotional satisfaction compared to the same responses they believe are from humans. The preference is so strong that participants are willing to wait for days or even weeks of human feedback rather than get AI assistance immediately.

“We’re entering an age where AI can produce responses that look and sound empathy,” said Anat Perry of the University of Jerusalem, a Hebrew University who leads an international research team. “But this study shows that even if AI can simulate empathy, people still want to feel what another person really understands, feels and cares about them.”

The authenticity effect

Researchers use complex deception to isolate the impact of perceived sources on transference assessment. Participants shared their personal emotional experiences and received AI-generated responses, but were told that these responses were from humans or chatbots. These responses are formulated by large language models, including cognitive understanding, emotional sharing, and real care.

The results are surprising. Artificial responses consistently earned higher ratings of empathy and support, while generating more positive and less negative emotions. The effect is maintained on different AI models, response lengths and even multiple conversations.

A key finding not highlighted in the press release: when participants even suspect that the “human” response may have been assisted, their positive feelings were greatly reduced. This shows that perceived authenticity – the belief that people really invest their time and emotional efforts plays a key role in our experience of empathy.

What are the most important aspects

Research shows that not all components of empathy are given equal importance. Emphasize the strongest preference for human resources:

• Emotional sharing (“feeling” with people)
• True care and attention (motivation empathy)
• Time and emotional effort
• True understanding, not just knowledge

When AI is prompted to provide only cognitive understanding (not expressing feelings or concerns emotions), the gap between human and AI preferences largely disappears. This shows that people can accept AI for analysis and advice, but desire human connections for emotional support.

Select an experiment

Perhaps most of the revealing are experiments where participants can choose between instant AI response or waiting for a large number of periods of human feedback. It is worth noting that many people choose to wait-not just for response, but even confirming that another person has just read their story.

Those who choose human interaction cite real understanding, emotional sharing, and the desires of those who truly care about. Meanwhile, AI chooses speed and convenience, or sometimes prefers AI to avoid potential judgments from others.

Impact on AI integration

These findings are of great significance for deploying AI in sensitive environments. While AI shows promise to expand support systems in healthcare, education, and mental health, research highlights critical limitations.

“AI can help expand support systems, but people still want the human feeling at times when deep emotional connections are needed,” Perry explained.

As AI aid becomes ubiquitous in communication (from email drafting to message editing), research suggests hidden costs. “The more we rely on AI, the more risk our words are, we feel empty,” Perry warned. “As people start assuming that every information is generated, perceived sincerity, with it, the emotional connection may begin to disappear.”

The future of emotional AI

This study raises fundamental questions about AI illuminating authenticity in the world. Although current AI lacks the ability to truly “with humans” or truly care about its health, it can provide valuable cognitive empathy and practical support without fatigue or bias.

Challenges are essential to know when it is crucial to have enough AI assistance. The study shows that AI is good at understanding and advice, and humans still cannot replace emotional sharing and real care.

As AI becomes more complex and human, these distinctions may become increasingly important and harder to sustain, fundamentally reshaping how we value and experience real human connections in an increasingly digital world.

Related