Transformer
The transformer is the design behind today's AI language models. Each word in a passage is compared with every other word to work out meaning.
The transformer is the design behind nearly every AI language model in use today. It's the T in GPT. Google researchers introduced it in a 2017 paper called Attention Is All You Need, and the models behind ChatGPT, Gemini, Claude and the rest are built on it.
You don't need to understand how it works to do well in AI search. It helps to know one thing it does, though, because it explains why AI platforms are good at meaning and less tied to exact keywords.
What it does
Earlier language models read text roughly in order, one word after another, and struggled to connect words that were far apart. A transformer looks at the whole passage at once. For every word, it works out which other words matter to it, a step called attention.
So in "Fernhill's subscription is cheaper than Kestrel's, but its beans are older," the model can work out that "its" means Fernhill and "older" describes Fernhill's beans, even though the words are some distance apart.
Why it matters for your content
Because the model reads for meaning, it can match a question to a page that uses different words. Someone asks for "a coffee that won't keep me up at night," and the model can connect that to a page about decaf without the page using that phrase.
That's good news and a warning. You don't need to stuff your pages with every way of saying something. But the model is reading what your sentences mean, so vague, muddled or contradictory writing comes through as exactly that.
An example
For example, Harrow & Pine's product page says "solid oak, hand-finished, seats six." A question about "a sturdy wooden table for a family of six" shares almost none of those words. A transformer-based model can still see that the page answers it.