When AI produces confident answers that turn out to be wrong, it's easy to call it a glitch. But most of the time, the problem isn't the model misbehaving, it's the information it learned from in the first place. Poorly sourced, unverified data creates systems that sound