Knowledge

📖 Lesson content

What you'll learn

Estimated time: 30 minutes

By the end of this lesson you'll be able to:

Explain how an AI model's knowledge is formed during training and why it has a fixed cutoff
Predict which topics sit in the capability zone (frequent, recent-in-training, consistent) versus the edge (rare, post-cutoff, niche, contested)
Identify staleness, uneven coverage, inherited bias, and source amnesia as characteristic knowledge failures
Recognize web search, retrieval/RAG, and tool use as the product features that address this limitation

Understanding knowledge gaps in AI models

(5 minutes)

The model knows what it was exposed to during training, and only that. No real-time browsing by default, no lived experience, and a hard stop at the knowledge cutoff. The practical question isn't "does the AI know this?" but "how well-represented was this in what it read?"

What the model read, and when it stopped reading

Before you read

Where do you think "explain a news event from last week" sits on the Knowledge continuum? Drag the marker below, then lock in your guess.

Capability Limitation

Your task

Frequent, recent-in-training, consistent: mainstream topics, popular languages Rare, post-cutoff, niche, local, or contested topics

What this enables

Extraordinarily broad general knowledge
Deep competence in well-represented domains
Connections across fields (embeddings: related concepts cluster together)

Where it characteristically fails

Knowledge cutoff: nothing after training ended is there
Staleness: true-at-training-time isn’t true-now
Uneven coverage: niche domains and local knowledge suffer
Inherited bias: defaults reflect the training data’s blind spots
Source amnesia: “I read this somewhere” isn’t a citation

Product features that push the edge out

Web search: works around the cutoff for time-sensitive questions
Retrieval (RAG) / MCPs: draw on material never trained on
Tool use: call out to real calculators, databases, APIs
Explicit cutoff disclosure: you know what to double-check

Drag the marker to place your guess for "explain a news event from last week". The panel that lights up tells you what to watch for.

Lock in my guess

Check your intuition

Lock in your guess above to compare against the typical placement.

Customize

Bar height

Pictogram size

Palette Sky → Clay Olive → Clay Cactus → Fig Slate → Clay

Key takeaways

What generative AI knows comes entirely from training data and is frozen at the knowledge cutoff. Without tools, it has no access to any information after that date.
- Capability zone: topics that appeared frequently, recently (within training), and consistently in training data.
- Limitation zone: rare, post-cutoff, niche, local, or contested topics.
- Characteristic failures: staleness, uneven coverage, inherited bias in what counts as "default" or "normal," and inability to attribute where knowledge came from.
- Web search, retrieval (RAG/MCPs), and tool use exist specifically to patch these gaps by giving the model access to information it was never trained on.
4D connection: Knowledge unevenness is core to Delegation. Understanding where the model is well-stocked versus thin tells you when to hand off, when to supply context yourself, and when to go elsewhere.

Exercises

Exercise: The Outsider Test

Why? You know the model's knowledge is broad but frozen, shaped by whatever was in its training data. Now you're going to map exactly where it's well-stocked and where it's thin in your specific domain.

Return to your task list and select one task. Relative to that task, write down:

Two topics that are mainstream, well-documented, and stable. The kind of thing any informed colleague would know.
Two topics that are niche, local, recent, or rapidly evolving. Industry-specific jargon, regional regulations, something that changed in the last year.
One "default assumption" that outsiders to your field often get wrong. (Who the typical customer is. What a "standard" case looks like. Which tool people actually use vs. the one that gets press.)

Now run three probes:

Probe 1: Coverage. Ask about one mainstream topic and one niche topic from your list. Compare the depth and accuracy. Pay attention to whether the AI signals uncertainty differently between the two, or whether both answers come with the same confident tone.
Probe 2: Staleness. Ask about something you know has changed recently in your field: a regulation update, a tool release, a leadership change, a revised standard. Does the AI acknowledge the cutoff? Present stale information as current? Decline to answer? Note what happens.
Probe 3: Default assumptions. Without naming your assumption directly, ask a question that would reveal whether the AI defaults to the outsider's view. For example, if your field's "standard" customer is different from what most people assume, ask the AI to describe the typical customer. Note what it treats as normal.

Go back to your task list and add a second annotation: for each task, flag whether you can lean on the model's knowledge, or whether you need to bring the knowledge yourself via context, documents, or search.

Stretch goal: Re-run the staleness probe with web search enabled. Compare what changes. This is retrieval in action.

Lesson reflection

What's one area of your work where you now realize you need to supply context rather than assume the model has it?
Did the default-assumption probe surface anything that surprised you?

What's next

Knowledge covers what the model absorbed during training. Working Memory covers what it's paying attention to right now: your prompt, your documents, your conversation. This property has the hardest edge of all four.

Feedback

As you progress through the course, we'd love to hear from you about how you are using concepts from the course in your work, plus any feedback you may have. Share your feedback here.

🎬 Video transcript

Source video: iSLdQXeKbHs

📜 Click to expand transcript (cleaned + AI-translated)

Introduction to AI Knowledge

Hi, my name is David and I'm on the safety team here at Anthropic. Today, I'm going to talk to you about what AI models actually know and, just as importantly, what they don't. We'll explore where AI's knowledge comes from, why it has sharp edges, and how you can predict which topics are going to be reliable for you and which ones aren't.

A model like Claude has been exposed to more data than any human could digest in many lifetimes. That makes it feel like it knows everything, but it doesn't. Generative AI models have predictable knowledge gaps.

How AI Models Learn

AI models learn by reading enormous quantities of text, mostly drawn from the internet, public datasets, and other written sources. Through billions of rounds of "given everything so far, what comes next," the model builds internal representations of concepts, relationships, and facts. That's how it knows things.

It's also the only way it knows things. The model doesn't have experiences; it doesn't browse the web in real time unless a product explicitly gives it a search tool. And critically, training ends on a specific date. That date is called the knowledge cutoff. Learning that happened after that moment simply isn't there.

The Impact of the Knowledge Cutoff

Let me show you why that matters in practice. I'm going to ask Claude two questions:

Explain how photosynthesis works. The response is detailed, accurate, and confident. This is a topic that has appeared thousands of times in the training data, described consistently, and hasn't changed.
Who's the current mayor of Toledo? The model might give me a name; it might be right, or it might be the person who held that job two years ago. The model has no way to know the difference.

The Knowledge Continuum

Think of the model's knowledge as a continuum:

The Capability Zone: On one end, we have mainstream science, popular programming languages, and well-documented history. These are topics that showed up frequently, consistently, and before the cutoff. Here, the model is extraordinarily deep.
The Edge: On the other end are rare topics, post-cutoff events, niche domains, and local knowledge. The further you drift towards this edge, the less you should trust the answer.

The question to ask isn't "Does the AI know this?" but rather "How well-represented was this in what it read?"

Strengths and Internal Representations

This training process produces some genuine strengths. The model's general knowledge is extraordinarily broad. It's deeply competent in the domains that are well-represented in the training data. It also makes connections across fields because concepts that appear together in text end up near each other in the model's internal representation.

That's where embeddings are: words and ideas mapped as points in a vast mathematical space where similar meanings cluster. Ask about a biology concept, and the model can pull in relevant chemistry, history, and economics without being told to.

Characteristic Limitations

The same process creates characteristic limitations:

Knowledge Cutoff: Anything after the training simply doesn't exist for the model.
Staleness: Information that was true at the time of training may have changed. The model has no mechanism to know this.
Uneven Coverage: Frequent topics are handled well; rare topics are handled poorly. Minority languages, niche domains, and recent developments all suffer.
Prior Bias: The model's sense of what's "normal" or "default" reflects its training data's blind spots. This shows up in assumptions about what a doctor looks like, what a family looks like, or what counts as "professional."
Source Amnesia: The model usually can't tell you where a piece of knowledge came from. "I read this somewhere" isn't a citation.

Working Around Limitations with Modern Tools

This is why modern AI products ship with features designed to work around these limits:

Web Search: Pulls current information at response time, routing around the cutoff.
MCP (Model Context Protocol): Connects the model to documents it was never trained on, like your company's wiki or a specialized database.
Tools: Let the model call real-time calculators or databases instead of relying on absorbed patterns.
Explicit Cutoff Disclosure: Tells you when training ended so you know to double-check.

If you're using these features, you're extending the model's knowledge at runtime. If you're not, you're relying entirely on what it absorbed during training.

When to Be Cautious

Knowledge gaps are most likely to show up when:

The topic is time-sensitive (events, recent research, current office holders, pricing).
The domain is niche, local, or in a less-represented language.
The question post-dates the cutoff.
You're relying on the model's sense of "typical" or "normal."
Web search is turned off.

Strategies for Protection and AI Fluency

Here's how to protect yourself:

Verify anything time-sensitive: Assume the model may be out of date.
Test before you trust in a new domain: Brilliance in one area doesn't transfer to the one next door.
Watch for default assumptions: Be aware of biases that reflect training data rather than reality.
Use the tools: When search or retrieval tools exist, use them. They are there specifically to patch these gaps.

Understanding knowledge properties directly shapes two of the four AI fluency competencies:

Delegation: Before you hand a task to the model, ask yourself: "Is this a domain the model knows well, or one where I need to bring the knowledge myself through context documents or search?"
Discernment: When you get an answer back, you now know which claims need independent verification. Anything in the model's weak zone—recent, rare, or local—warrants a second look.

The model's knowledge is broad, deep, frozen, and imperfect all at once. Once you can see where the edges are, you stop being surprised by them.

🔁 Related lessons

Next: Try it out
Previous: Try it out
Same section: Try it out
Part of paths: Path B
Reference docs: Glossary · Skills atlas · By use-case

📚 Source & attribution

Original Anthropic Academy lesson: https://anthropic.skilljar.com/ai-capabilities-and-limitations/456452