Hallucination
A language model output that is fluent and plausible but factually incorrect, fabricated, or unsupported by source material, occurring when the model generates content based on training-data patterns rather than grounded evidence.
How it works
Hallucination is the first failure mode enterprise buyers worry about, and rightly. A confident, well-formatted answer that is wrong is more dangerous than an obviously broken one. The mitigation toolkit is well understood: ground outputs in retrieved evidence (RAG), require citations to source documents, enforce structured output formats, validate factual claims against authoritative data, and use abstention prompting (allow the model to say "I do not know"). For UK regulated workloads (FCA-regulated, SRA-regulated, NHS clinical), hallucination is treated as a compliance issue rather than a UX issue: the architecture must ensure that any factual claim has a verifiable source, and the system must refuse to answer rather than guess. Ayoob AI builds production systems on this principle.
Related terms
Retrieval-Augmented Generation (RAG)
An architecture pattern that grounds language model outputs in retrieved documents from a private corpus, reducing hallucination and enabling answers based on the firm's own data rather than the model's training set.
Prompt Engineering
The discipline of designing, structuring, and refining the input text passed to a language model to produce reliable, accurate, and properly-formatted output for a specific task.
Human-in-the-Loop (HITL)
A design pattern in which humans review and approve AI decisions at designated points in a workflow, used for high-stakes or low-confidence cases where full automation is inappropriate.
Want to see this technology in action?
Book a Discovery Call