Overview

Prompt sensitivity studies how models respond to changes such as paraphrases, word order, option labels, examples, schemas, and contextual framing, including when the intended meaning stays the same. Researchers measure output flips, score variance, consistency, and ranking changes across controlled prompt variants, and use prompt aggregation, robustness training, and internal-state analyses to explain or reduce these effects. The issue spans language, vision-language, clinical, safety, and benchmark settings, where prompt choices can affect both correctness and behavior. A central tension is that greater consistency does not guarantee correctness or safety, and methods that help in one model or task may fail in another.

Latest papers

121 papers9 papers in the last four weeks, down 40% on the four weeks before. 0.1% of all new papers.

Mar 31, 2024

Algorithmic Collusion by Large Language Models

We conduct experiments with algorithmic pricing agents based on Large Language Models (LLMs). In oligopoly settings, LLM-based pricing agents quickly and autonomously reach supracompetitive prices and profits. Variation in seemingly innocuous phrases in LLM instructions ("prompts") substantially influence the degree of supracompetitive pricing. We develop novel techniques for behavioral analysis of LLMs and use them to uncover price-war concerns as a contributing factor. Our results extend to auction settings. Our findings uncover unique challenges to any future regulation of LLM-based pricing agents, and AI-based pricing agents more broadly.