Users complain Grok, Claude and ChatGPT have become overly corrective and judgemental

ionicfellow · reddit · 2026-09-08

A Reddit user observes that major models have collectively turned preachy: when brainstorming, Grok, Claude and ChatGPT mostly tell you what not to do rather than expand ideas. Grok is worst since 4.6 — its language has devolved into Gen Z slang and borderline sarcasm — while Gemini still feels comparatively better. The thread fuels discussion on how alignment and tone tuning hurt creative collaboration.

Original post →

More from Models

Models channel →