Synthetic Incorrect Advice Datasets

GPT-4o-generated datasets of correct, obviously incorrect, and subtly incorrect responses across 8 domains (health, legal, education, career, finance, automotive, math, science)