3 ms·
This was a good and succinct read. I noticed this phenomenon a lot when googling for non-political phrases. It's really all over Google search. Phrasing a sea
by moritzwarhier 4y ago
This was a good and succinct read.
I noticed this phenomenon a lot when googling for non-political phrases. It's really all over Google search.
Phrasing a search in a way that implies some information or judgement tends to confirm the presupposition, even it has been proven to be wrong.
This would be expected to some degree even with plain keyword matching.
But AI and knowledge graph enhancement seem to amplify the problem.
- jklm 4y agoI’m no fan of Google in some ways, but is this a problem specific to search or Google search? I don’t see how you can avoid this unless you proactively filter out wrong information…which leads to a question of who decides what’s right. e.g. Should masks be rightfully mandated in classrooms? As engineers, I feel like we bias for clear cut conclusions but the world is rarely black and white.
- moritzwarhier 4y agoYes, I think it's specific to Google but only to a degree. Google provides the most richely "enhanced" search experience, also but not only when using natural language queries. I think you are right about the "Who decides what's right/correct" part. But I thought the article's argument was kind of similar: There are no clear-cut solutions, so search engines shouldn't pretend to be able to provide them. Google often gives factually wrong or subjective "answers" to queries that use a natural-language question format, and presents them as fact. Google's own AI-generated "Q&A" snippets and "Knowledge Graph" answers are presented as somewhat equivalent to the fact snippets that Google pulls from Wikipedia. It is this kind of highlighted "knowledge answers" to arbitrary queries that the article talks about, I think.