RAG & Search
Learning to rank
Teaching AI to put items in a useful order, such as the most relevant search results first.
Example
A search service learns which results should appear higher for a query.
Why people use it
It teaches a search or recommendation service which items to put first.
What you'll hear
“Which results are most useful for this question?”
What this means for you
Ask what outcome the ranker was trained to prioritize.
Can you control it?
Developer-only
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Can people supply the ratings used to teach ranking?
- Yes. Human judgments can guide the order, alongside other signals such as clicks or task results.
- Does the first item have to be the most accurate?
- No. The service may be trained to favor other things, such as clicks or purchases.
- Can people clicking the first result influence future rankings?
- Yes. Earlier placement can create more clicks, which may then affect what the system learns.