Skip to main content
Core concept · AI Search Infrastructure

Definition

Lost in the middle is a documented tendency of large language models to use information at the beginning and end of a long context window well, while under-weighting information placed in the middle. The same fact influences an answer more or less depending on where it sits in the context the model is given. Retrieval places sources into a context window, and position affects whether they get used. A passage retrieved but buried in the middle of a long context can be effectively ignored, so being retrieved does not guarantee being read. This raises the value of concise, high-signal passages that earn a top position, and it is part of why shorter, denser sources often outperform longer ones in AI answers. The behavior belongs to the models, so the response is to be the source worth placing first.

Context window

See also

RAG

See also

Reranking

See also

Passage ranking

See also

Chunking

See also

Relevant Plate Lunch Collective Services

Citation-Ready Content AI SEO