Understanding Claude AI – What It Is and How It Thinks Log in and enroll to track lesson completion. By the end of this lesson, students should be able to: What kind of tool is Claude? Claude is a large language model (LLM) – a type of AI system trained on a very large collection of text from books, websites, articles, code, and other written sources. During training, it learned statistical patterns about how words, sentences, ideas, and concepts relate to each other. It did not memorize a database of facts. It learned how language works. This distinction matters enormously for how you use it. When Claude answers a question, it is not querying a database the way a search engine does. It is generating a response token by token, based on what patterns in its training suggest is a plausible, coherent, helpful continuation of your input. This makes Claude excellent at reasoning, writing, summarizing, explaining, structuring, and drafting – and it makes Claude unreliable as a standalone source of verified facts. Claude is a thinking tool, not a fact database. Think of Claude less like an encyclopedia and more like a highly experienced colleague who has read an enormous amount – but who you still need to double-check when facts matter. That colleague can help you think through a problem, draft a document, spot weaknesses in your plan, suggest approaches you hadn't considered, and explain complex topics clearly. But you would not cite that colleague in a legal filing without verifying their claims independently. Claude does not browse the internet in the default interface. Unless you are using a version of Claude that has been explicitly connected to a live search tool, Claude cannot look things up in real time. It works from what it learned during training, which has a knowledge cutoff date. Anything that happened after that date is not in its training data. Claude does not have memory between conversations by default. Each new conversation starts fresh. Claude does not remember what you discussed yesterday, last week, or in a previous session unless you paste that context back in, or unless you are using a product that has added memory features. This is important to understand because students often assume Claude is building a relationship with them over time – by default, it is not. Claude is not conscious, opinionated in a personal sense, or trying to please you. Claude is designed to be helpful, harmless, and honest. It will push back on requests it considers harmful or misleading. It will acknowledge uncertainty when it has it. It is not telling you what you want to hear – it is generating what it calculates to be a useful response. Understanding this helps you interpret Claude's answers more accurately. A student asks Claude to help them research a competitor's pricing strategy. Weak prompt What is [Company X]'s current pricing? This treats Claude like a search engine. Claude may produce a plausible-sounding answer based on pricing it saw during training – which could be months or years out of date, or could be a hallucinated estimate. Improved prompt I'm researching how SaaS companies in the project management space structure their pricing tiers. I don't need live data – I need help thinking through the common patterns (per-seat, flat-rate, usage-based, freemium) and what tradeoffs each creates for customer acquisition vs. revenue predictability. Help me build a comparison framework I can use when I look up current pricing myself. The improved prompt uses Claude's actual strength – pattern recognition, framework building, analytical reasoning – rather than asking it to retrieve current facts it may not have. Claude performs better when you tell it what role to reason from. Prefacing a prompt with "You are acting as a strategic advisor helping a small business owner evaluate pricing models" gives Claude a useful frame that shapes the vocabulary, depth, and angle of its response. This is not a trick – it is giving Claude the context it needs to aim at the right target. Verify in your current Claude interface whether persistent system prompts are available, as this feature may vary by product tier. Claude can produce confident-sounding answers that are factually wrong. This is called hallucination, and it happens because Claude is generating plausible text, not retrieving verified facts. The risk is highest when asking about specific numbers, dates, names, citations, legal or medical specifics, or anything that happened recently. Never use Claude's output as a final source for decisions that carry real consequences without independent verification. Write a two-sentence explanation of what Claude is, in your own words, as if you were explaining it to a coworker who has never used it. Then write one example of a question where Claude's answer should always be verified before acting on it. After completing this lesson, you should be able to explain to someone else what Claude is, why it is not a search engine, and give one example of when you would and would not trust its output without verification. Log in and enroll to take this lesson quiz.
Lesson 1: What Claude Is – and What It Is Not
Lesson Objectives
Lesson Content
Practical Example
Lesser-Known Tip
Safety Notes
Practice Task
Completion Check
Lesson Quiz