Can we really trust reasoning - Pretrained | Wave AI Podcast Notes