alleryStart learning →What is the main reason AI sometimes produces confident but wrong answers?
AIt runs out of memory mid-sentence
BAI predicts plausible text, no truth-check step
CIt only reads the first line of a question
DIts vocabulary list is incomplete
Answer & Solution
Correct answer: B. AI predicts plausible text, no truth-check step
**Card 1 and Card 2:** AI training fits patterns from text.
The model picks the highest-probability continuation.
There is no final truth-check step (Claude's Constitution, p.6: 'AI training is still far from perfect').
Options A, C, D describe unrelated mechanisms.
Keep practising AI Basics
Browse all AI Basics questions for Claude AI, or explore more Claudearchitect topics.
Related questions
Pop the jobs a model does wellA friend asks: what should you do first when an AI answer contains a specific date?Which type of claim carries the highest hallucination risk?AI says an essay has 850 words. The real count is 1003. What does this show?What does AI do well when you ask it to compare two options?Which task is AI best suited for?Why is AI especially strong at language tasks like summarising and rewriting?During fine-tuning, what do human raters actually do?