Anthropic staff reportedly as confused as users by Claude's output

sanderssays · x · 2026-09-01

A user questioned whether Anthropic employees actually understand Claude's output or are just as confused as the rest of us. Comments highlighted that Claude's responses are often filled with empty buzzwords like "Sharp," "Defensible," and "Cleaner," dismissing it as "cute gibberish" and reflecting the common issue of flowery language in LLM outputs.

Original post →

More from Models

Models channel →