🚀NEW LABGetting Started with Claude AgentsStart lab
Reasoning · Evaluation

EmotionPrompt

First page
EmotionPrompt
Paper summary

Microsoft researchers show that appending emotional stimuli to prompts reliably improves LLM performance across 45 tasks.

Ask this paper

Key points
01

45-task evaluation: Tested across Flan-T5-Large, Vicuna, Llama 2, BLOOM, ChatGPT, and GPT-4 on 45 deterministic and generative tasks.

02

Emotional stimuli: Appends phrases like "This is very important to my career" to prompts, drawing on social-psychology theories of human motivation.

03

Consistent gains: Produces consistent improvements across both smaller and frontier models, despite the prompts being content-free manipulations.

04

Emotional-intelligence signal: Suggests LLMs have internalized patterns connecting emotional framing to effort - a "bug or feature" question that has driven follow-up research on LLM behavioral psychology.

Every Monday
Get next week’s papers.
Subscribe on Substack