User Rants: Claude Now Returns Paragraphs That Answer Nothing
AccidentalFolklore · reddit · 2026-11-20
A Reddit user with brainstem compression and ADHD complains that Claude (worse above 4.6) and ChatGPT now return paragraphs of text that don't answer the question — "the shrug is me shorthanding the five paragraphs it used to explain nothing." Concrete cases: checking data loss across directories, debugging Obsidian sync writing zeros to files, fixing Rust/Tauri wiped by a work-network update — the official models go in circles while a jailbroken variant instantly gives the five lines of code that would have fixed it six prompts earlier. Handing off whole tasks works fine, but conversational troubleshooting has become a nightmare, and their productivity has dropped noticeably.
More from Models
- Gemini API previews tool retrieval: schemas load on demand, cutting token cost — petrusenko_max · 2026-11-20
- "Neon gave me AI psychosis": researcher describes overconfidence spike from new top model — andrew_n_carr · 2026-11-20
- "Beating Gemini is the benchmark now": CursorBench chart mocked as benchmaxxing — zacharynado · 2026-11-20
- repligate: no Anthropic ban on AI welfare research — refusals likely mean you're skeeving Claude out — repligate · 2026-11-20
- Artificial Analysis launches Data Playground for building custom weighted benchmark indexes — ArtificialAnlys · 2026-11-20
- TBPN: Google 'wasn't AI-pilled enough' — selling TPUs vs. scaling tension — zacharynado · 2026-11-20