How AI works
Learn how AI models work.
Learning the basics
What happens when you talk to AI?
When you send a message to an AI, there's a moment where it appears to be “thinking.” But what's actually happening? This video walks you through what's going on under the hood.
Can you trust what AI tells you?
Ask an AI a question and the answer comes back confident, well-organized, maybe even cited. Can you trust it?
What does AI know about me?
People sometimes ask AI things they wouldn't broadcast, like medical questions, money worries, and emails they're nervous to send. So when you give it details, what does it actually know?
Capabilities and limitations
AI capabilities and limitations
Build an accurate mental model of what large language models can and cannot do: next-token prediction, knowledge, working memory, steerability, and context limits.
The 4 Properties of AI
This tutorial offers a full overview of the 4 properties of AI. It's a quick reference to help you understand the things that make AI capable in some situations and limited in others.
Why do AI models hallucinate?
Learn what AI researchers mean when they talk about hallucination in AI models, why it may occur, and tactics you can use to spot this in your conversations.
What is sycophancy in AI models?
Learn what AI researchers mean when they talk about sycophancy, when it's more likely to show up in conversations, and tactics you can use to steer AI towards truth.
Why does bias exist in AI models?
Learn what bias is, why it may occur, and tactics you can use to spot bias in your AI conversations.
Going deeper
Tokens: why some inputs cost more than others
Claude usage is measured in tokens. Whether you're monitoring a usage bill or wondering when a rate limit will kick in, tokens are the unit of measurement. Learn what a token is, predict how many a piece of text will use, and see how tokens encode meaning.
How context affects Claude's performance and cost
Everything Claude knows about the world was either baked into the model during training, or is part of the current context. Learn where Claude's answers come from, where you might be using more tokens than expected, and how to be intentional about the tradeoff between cost and quality.