A new research paper explores methods to improve the alignment between human and machine understanding of psychological constructs using large language models (LLMs). The study empirically assesses various prompt engineering strategies, including codebook-guided selection, automatic generation, persona prompting, chain-of-thought, and explanatory prompting. Findings indicate that the most effective approach involves a few-shot prompt combining codebook-guided empirical selection with automatic prompt engineering, emphasizing the importance of clear construct definitions and task framing. AI
IMPACT This research offers a systematic method for optimizing LLM prompts in specialized domains where precise alignment with expert judgment is critical.
RANK_REASON Research paper published on arXiv detailing empirical assessment of prompt engineering for LLMs. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
