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.Why It Matters for AI Search
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.Related Terms
Context window
See also
RAG
See also
Reranking
See also
Passage ranking
See also
Chunking
See also