← The notebook

AI & the internet

Why AI recommendations aren't objective

A recommendation is an answer shaped by what a system can see.

Sample draft for this website preview. This is not a previously published article by Resync.

A question worth keeping open

When an AI recommends a business, what are we actually looking at? A judgment of quality, a reflection of available information, or a little of both?

Working on GEO tools makes this a useful question to investigate. The answer a user sees is only the surface. Underneath are retrieval choices, missing context and an evaluation process we may not be able to inspect.

What I want to test

A useful next step would be to ask the same question across different systems, record the sources, and separate what was retrieved from what was ultimately recommended.

Being mentioned is one signal. It is not the whole story.

More from the notebook ↗