Research questions LLM empathy control reliability

A new research paper published on arXiv questions the reliability of controlling large language models (LLMs) for empathy. The study tested three instruction-tuned LLMs, including Qwen, Llama, and Gemma, using automated empathy scores and a discriminative classifier. While interventions could significantly alter the text and partially shift affective empathy scores in models like Qwen, the researchers found that detection of empathy directions does not necessarily equate to reliable control, especially for cognitive empathy, which proved difficult to measure accurately. AI

IMPACT Raises questions about the reliability of current methods for controlling LLM behavior, particularly in nuanced areas like empathy.

RANK_REASON Academic paper published on arXiv detailing research findings. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• EPITOME
• Gemma
• Gotit.pub
• Hugging Face
• Llama
• Qwen
• ScienceCast

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →