# Hallucination

Concept · [Building with AI](https://ecosystem.samcarlton.com/ai)

Also called: hallucinate, hallucinations, made-up answer, confabulation

When an [AI model](https://ecosystem.samcarlton.com/model) states something false or invented as though it were true.

**For example:** You ask an AI tool for a court case that supports your argument. It gives a convincing case name that does not exist.

## Connects to (20)

- [LLM](https://ecosystem.samcarlton.com/llm): Large language model.
- [RAG](https://ecosystem.samcarlton.com/rag): Retrieval-augmented generation.
- [Prompt](https://ecosystem.samcarlton.com/prompt): The instructions or question you give an AI tool in everyday words.
- [Parametric knowledge](https://ecosystem.samcarlton.com/parametric-knowledge): What a model knows from its training, stored in its parameters.
- [Sycophancy](https://ecosystem.samcarlton.com/sycophancy): The habit AI models have of telling you what you want to hear: agreeing too readily, praising weak ideas or giving up a correct answer…
- [Evals](https://ecosystem.samcarlton.com/evals): Tests that check whether the outputs of a model or an AI product meet the criteria its builders set.
- [Knowledge cutoff](https://ecosystem.samcarlton.com/knowledge-cutoff): The date where a model's training data ends.
- [Next-token prediction](https://ecosystem.samcarlton.com/next-token-prediction): The one job a language model performs.
- [Smart zone](https://ecosystem.samcarlton.com/smart-zone): A name for the early part of a session, when an agent is focused and accurate.
- [Contextual knowledge](https://ecosystem.samcarlton.com/contextual-knowledge): Facts the model can read directly in front of it right now, such as documents you pasted, files the agent opened or search results.
- [Secondary source](https://ecosystem.samcarlton.com/secondary-source): A description of the original material rather than the material itself, such as a summary, a README or a compaction note.
- [Probabilistic vs deterministic](https://ecosystem.samcarlton.com/probabilistic-vs-deterministic): Deterministic software gives the same output for the same input every time.
- [AI literacy](https://ecosystem.samcarlton.com/ai-literacy): Enough understanding of AI to use it well and safely: what it is good and bad at, how to write a prompt, how to check its work and what…
- [Ground truth](https://ecosystem.samcarlton.com/ground-truth): The verified, correct answers or facts that an AI system's output is checked against.
- [Human in the loop](https://ecosystem.samcarlton.com/human-in-the-loop): Designing an AI process so a person reviews, approves or corrects the AI's work at key steps instead of letting it act alone.
- [Temperature](https://ecosystem.samcarlton.com/temperature): A setting that controls how random an AI model's word choices are.
- [Perplexity](https://ecosystem.samcarlton.com/perplexity): An AI search tool.
- [Gemini Notebook](https://ecosystem.samcarlton.com/gemini-notebook): A Google research tool, formerly called NotebookLM, that answers questions using the documents a person adds, with citations, and can turn…
- [AI Overviews](https://ecosystem.samcarlton.com/ai-overviews): AI-written summaries that Google Search shows at the top of some results, with links to the pages they draw on.
- [Blue J](https://ecosystem.samcarlton.com/blue-j): An AI research tool for tax professionals.

Map view: https://ecosystem.samcarlton.com/hallucination