How AI Pays Attention
AI doesn't treat all words equally. Every word scores every other word for relevance, then blends their meanings using those scores — resolving ambiguity and tracking context across the whole sentence.
Two kinds of bank
Same word, same number, same spot on the map.
Yet you read two different things. How does the AI?
Words look back
Before deciding, every word looks back at the other words in the sentence.
The pull
The words it pays attention to pull its meaning their way.
You be attention
You be the attention. Tap the word that tells you which bank this is.
Back to your sentence
Back to your sentence. The last “the” looks back too.
It leans on “sat on”. Now it carries the idea that something you sit on comes next.
Words look around
Every word looks at the others and borrows meaning from them.
This is called attention. Real models do it many times over, in many layers.