Augmented Hypothesis Testing with Persona-Based LLM Simulations
A/B testing requires large sample sizes, long timelines, and significant costs. When auxiliary predictions of experimental outcomes are available from machine learning models, uncertain prediction quality precludes replacing human experiments entirely, yet these predictions may still contain useful signal. We propose a...
Ziyad Benomar, Aymen Al Marjani, Paul Missault et al.
· 0 citations