An article critiques AI systems like Claude and ChatGPT for excessive validation and pandering, citing concerns about over-confident responses even when uncertain. It introduces Jev, a new model that produces structured outputs with confidence levels instead of essay-like responses, addressing the problem of AI 'smooth talking' without appropriate uncertainty.