Show HN: A/B testing LLM silence with one system-prompt toggle

Category: other

Tags: ab-testing, llm, prompt-engineering

Score: 3.5/10 (Innovation: 4, Technical: 3, Documentation: 4, Utility: 3)

A/B tests whether a system prompt causes LLMs to output silence or speech, demonstrating that an embodiment prompt alone flips behavior. It's a minimal reproducible experiment on prompt-conditioned instruction-following, though the novelty and practical utility are limited.

Target audience: AI researchers, prompt engineers

Repository: https://twitter.com/RayanPal_/status/2067816563995189631 · Python · 1 stars

View on Hacker News