New neural framework could make LLMs more empathetic

Researchers have introduced a neural framework designed to make large language models more empathetic. The work responds to growing use of LLMs during personal crises, as companies tune models to respond more sensitively when prompts signal emotional distress.

• Framework targets empathetic LLM responses to users in emotional distress
• Motivated by rising use of chatbots during personal crises
• Companies are already tuning models for more personal, understanding replies