Lv.1 0 XP

How Claude Actually Works — The Honest Explanation

Foundations 8 min +20 XP
💡
THE ANALOGY

Claude works like a very sophisticated autocomplete. When you start typing 'I feel like having...' your phone might suggest 'pizza' or 'a nap.' Claude does this too, but at a scale that produces coherent essays, detailed plans, and nuanced answers. It's predicting what words should come next, based on everything it has ever read.

⚠️ EXAM TRAP — The Wrong Answer People Choose

Claude does not 'look things up' when you ask a question. It generates answers from patterns in its training. This is why it can be confidently wrong — it patterns-matches, not fact-checks. Always verify facts that matter.

KEY POINTS
1 Claude was trained on a massive amount of text — it learned patterns of language, facts, and reasoning from that text.
2 When you ask Claude something, it generates a response word by word, predicting what should come next based on its training.
3 Claude has a 'knowledge cutoff' — it doesn't know about events that happened after its training ended.
4 Claude cannot search the internet by default — it works from what it already knows.
5 Claude has a context window — it can only remember what's in your current conversation. It forgets everything when you start a new chat.