Measuring AI search is hard because answers are non-deterministic and shift...

https://iris-wiki.win/index.php/Why_Prompt_Quirks_Change_Mention_Extraction_Reliability

Measuring AI search is hard because answers are non-deterministic and shift wildly due to geo variability. Off-the-shelf tools can’t track this volatility. That’s why we’re building our own stacks