Hallucination
In one sentence
When an AI presents non-existent or incorrect information with full confidence, such as fabricating data, sources, legal clauses, or product features.
Hallucinations occur because models generate text that looks fluent rather than verifying facts. If information is missing from the reference material, the model may simply fill it in based on common patterns.
Best practices to reduce hallucinations include restricting the AI to retrieve facts solely from provided sources (knowledge bases, RAG), requiring it to cite sources, verifying numbers, dates, and legal clauses manually or programmatically, and explicitly instructing it to state 'no data available' when information is absent.
Example
Asking an AI to write an industry article and getting 'a 2025 report from an agency shows 37% market growth,' only to find out upon fact-checking that the report does not exist.
Related terms
- RAG (Retrieval-Augmented Generation)
- A method where AI retrieves relevant information from a knowledge base first, then generates an answer based on those materials.
- Knowledge Base
- A repository of company documents and institutional knowledge that AI can search and reference, enabling it to answer questions and produce content grounded in internal facts.
- Temperature
- A parameter that controls the randomness of an LLM's responses. Lower values produce more stable answers, while higher values lead to more varied outputs.
- Large Language Model (LLM)
- AI models trained on massive volumes of text that can understand and generate text. Doubao, DeepSeek, ChatGPT, and Claude are all powered by large language models.
