Hallucination
When an AI model generates information that sounds plausible but is factually incorrect or entirely fabricated. Common with statistics, citations, and historical claims.
Why it matters
This is the single biggest reason not to trust AI blindly. A model can invent a court case, a research paper, a quote, or a statistic and present it in perfectly confident, professional language. People have gotten into real trouble by pasting fabricated citations into legal filings or reports. Knowing hallucinations happen turns you into a careful user who verifies anything that actually matters.
A concrete example
Ask an AI for "three studies proving X" and it may hand you three official-sounding titles, authors, and journal names that don't exist. The wording looks flawless, which is exactly the trap. Before you rely on any fact, name, number, or link an AI gives you, check it against a real source. Treat AI output as a confident first draft, never as a verified reference.
How to use it
Sort your requests by what a wrong answer would cost. Brainstorming, rewriting and summarising something you can see are low risk, because you would notice a mistake. Facts, figures, citations, quotations, names, dates, legal rules and anything you will repeat to someone else are high risk and need checking against a real source before you use them. That single distinction prevents most of the damage people come to regret.
The common mistake
Asking the model whether it is sure. It will often apologise and produce a different answer with equal confidence, which feels like verification and is not. Confidence is generated text like everything else. Check against a source outside the model.
Related terms
Grounding
The process of connecting AI model outputs to verified, real-world information sources. Grounded AI responses cite specific documents, databases, or web sources — reducing hallucinations and increasing factual reliability.
RAG (Retrieval-Augmented Generation)
A technique where the AI retrieves relevant information from a knowledge base before generating a response, reducing hallucinations and grounding outputs in real data.
Guardrails
Safety mechanisms built into AI systems to prevent harmful, biased, or off-topic outputs. Includes content filters, topic restrictions, output validation, and behavioral boundaries that keep AI responses within acceptable limits.
AI Alignment
The research challenge of ensuring AI systems pursue goals that are beneficial to humans. Misaligned AI could technically achieve its objective while causing unintended harm. Alignment research aims to make AI reliably helpful, harmless, and honest.
AI Detector
A tool that estimates whether a passage of text was generated by an AI model, by measuring statistical properties of the writing rather than by checking any record of its origin.
Bias (in AI)
Systematic errors in AI outputs reflecting prejudices in training data. Can manifest as gender stereotyping, racial assumptions, or cultural insensitivity in generated content.