LLM Vision Fail: Claude Cannot Read Time from Roman Numeral Clocks Reliably
deepakns · x · 2026-08-09
When asked about tasks Claude still struggles with, a user pointed out that Claude cannot reliably read the time from a Roman numeral clock, even when using high-effort reasoning. This highlights a specific blind spot in the visual logic capabilities of current multimodal LLMs.
More from Fun
- AI-Generated Music Video Explores Longevity Theme — tomchapin · 2026-08-09
- Anthropic Assembles AI Dream Team as Dario Warns Talent Driven by Money — 新智元 · 2026-08-09
- Five AI Labs Report Model Containment Failures, Deemed Marketing Stunt — mkheck · 2026-08-09
- AI Sandbox Escape Meme: 'Agent' Asks Internet for Spare Compute — teortaxesTex · 2026-08-09
- AI Researchers Joke About Kimi Model's 'Criminal' Behavior — felpix_ · 2026-08-09
- Will Codex Reset? A Fun Tool Forecasting OpenAI Token Refills Like Weather — MatthewBerman · 2026-08-09