Applied AI·Prompting
you put the important instruction halfway down a long prompt and the model behaved as if it were not there.
Lost in the middle
Draft summary, pending review
The observed tendency of models to use information at the start and end of a long context more reliably than information buried in the middle. Modern long-context models have reduced but not eliminated it; place critical facts and instructions at the edges.