unicornAll work

Knowledge & hallucinations

What a model actually knows, the training cut-off, and why it invents facts confidently — with the fixes.

In a nutshell Property #2 of four: what a model actually knows — and why it can be confidently wrong. Its knowledge is "baked in" during training, goes stale at a cut-off date, and doesn't cover everything. And since it builds a plausible continuation (see next-token prediction), on the unfamiliar it doesn't go quiet — it smoothly makes things up. That's a hallucination.

Where the model's knowledge lives

During training the model tunes billions of weights to predict text well. Facts and patterns from what it read are dissolved into those weights — this is parametric knowledge. Three consequences:

Analogy. An erudite locked in a room with no internet and no clock. They remember enormously much — but only up to the day they were shut in, and only what they'd read. Yesterday's news or your private document, they cannot know.

Why a model can be "confidently wrong"

Careful The model has no built-in sense of "I don't know this". Its job is to continue text plausibly. To a question with no answer in its knowledge it won't, by default, say "I don't know" — it builds the most plausible-sounding answer. Smooth, confident and invented.

That's a hallucination — a falsehood in the same confident tone as the truth. Typical victims:

Why: to a token predictor, a "plausible link" and a "real link" look equally good — both continue the text smoothly. It can't tell them apart without an external check.

How to work with it

Hallucinations aren't a bug that gets fixed — they follow from the model's nature. So you work around them:

  1. Put the facts in the prompt. Not "what does this law say?", but paste the law and ask about it. The model is great with what's in front of it (see working memory & context). This is the RAG principle — "add the source, then ask".
  2. Give it tools. Search, database/document access, a calculator — then facts come from a real source, not from memory.
  3. Ask it to cite and admit uncertainty. "Answer only from the given text", "if unsure, say so", "state where the fact comes from".
  4. Double-check the critical yourself. Numbers, dates, quotes, legal/medical facts, links — always against the primary source. A confident tone is no guarantee.
  5. Mind the cut-off. Recent events — only via search/data, not "off the top of its head".

Takeaways

Source

Anthropic Academy course AI Capabilities and Limitations, section "Knowledge".

All theory