You ask Claude about a very obscure library you just wrote. It gives a plausible-sounding but wrong answer. Why?
AFine-tuning was skipped for coding questions
BThat library is not in Claude's training data
CClaude only reads English-language text
DPretraining was not run long enough
Answer & Solution
Correct answer: B. That library is not in Claude's training data
Card 5 (try-it card) of this block: Claude answers confidently when a concept appears thousands of times in its training data; for a brand-new or private library, the information simply does not exist in that data so the model guesses from patterns. Source: simple_wikipedia_ai_ocr.md (AI learns from data) + claude_constitution_ocr.md (training data shapes Claude's knowledge).
Related questions
A friend asks: what should you do first when an AI answer contains a specific date?Which type of claim carries the highest hallucination risk?What is the main reason AI sometimes produces confident but wrong answers?You told AI your name yesterday. In a new chat today, AI does not know it. Why?AI says an essay has 850 words. The real count is 1003. What does this show?You ask AI for today's cricket score. Why can't it answer reliably?What does AI do well when you ask it to compare two options?Which task is AI best suited for?