AI chatbots can be as effective as humans at emotional support

Led by the University of Manchester, the study compared messages of emotional support generated by large language models (LLMs) and humans in everyday non-clinical contexts.

Emotional support

Dr Sarah Walker, from our School of Education, is a co-author of the research, which comprised five studies involving a total of 1,233 participants.

Across the studies, the team compared emotional support messages generated by three leading LLMs (ChatGPT 4o, Claude 3.5 Sonnet, and Gemini 1.5 Pro) with those written by human participants.

These responses were based on everyday emotionally charged scenarios, such as being passed over for a promotion or coming home to find a shared house untidy.

Participants were not told if the message they received came from a human or an AI, although many correctly identified the source of the messages.

Emotional context

The research team found LLM-generated responses largely mirrored strategies used by humans to manage others’ emotions. ChatGPT showed the highest similarity with human patterns.

In situations involving anger and fear, LLM responses were seen as more emotionally supportive than those written by people. They also helped participants feel more positive, for example calmer or happier.

In scenarios involving sadness, however, LLM and human responses were found to be equally effective.

Simply acknowledging someone’s feelings (e.g., “I totally understand why you feel sad”) did not fully explain why LLM responses were sometimes more effective. Adding or removing this emotional validation did not make a difference to how helpful the messages were perceived to be.

The researchers found, however, that the effectiveness depends on the emotional context and the key ingredient: specific, actionable guidance.

These are clear practical suggestions that people could use (such as breathing techniques, step-by-step reframing, or practical next steps). These specific suggestions improved emotional outcomes regardless of whether the message came from a human or an AI.

Find out more

• Read the full research paper A Digital Shoulder to Cry On: Understanding Why Large Language Models Can Be Effective in Extrinsic Interpersonal Emotion Regulation published in the journal Emotion.
• Find out more about Dr Sarah Walker.
• Our School of Education is ranked joint 66^th in the QS World University Rankings by Subject 2026 and third in the UK in the Complete University Guide 2027. Visit our Education webpages for more information on our undergraduate and postgraduate programmes.