ReportSSRNMar 5, 2025
Prompting Science Report 1: Prompt Engineering is Complicated and Contingent
Lennart Meincke, Ethan R. Mollick, Lilach Mollick, Dan Shapiro
Read the original →What follows is our summary, written for district staff. It is not the paper. If this is going to inform a decision, read the original.
Prompting Science Report 1: Prompt Engineering is Complicated and Contingent
Article Summary
This report from March 2025 indicates that prompt engineering techniques and approaches have inconsistent results. Specifically, politeness, once thought to improve LLM outputs, was shown to sometimes reduce output quality. These results emerged from repeated trials with identical prompts; for an individual user, evaluating the quality of each individual output remains key.
Related across the site
ResearchGetting started with AI: Good enough promptingEthan Mollick argues that using AI well takes no prompt engineering, just about ten hours of experimenting to learn w…ResearchAI Tutoring Outperforms Active LearningAn AI tutor built on pedagogy best practices helped students learn more, in less time, than an active, instructor-led…