Yep, I've also noticed a lot of really weird stuff from sketchy sources embedded in the answers panel.
We forgot that the reason search engines were effectively exempt from libel/copyright etc. is because were just describing the web, they published the map but the content was on all the individual websites.
It's a lot harder to argue that model still applies when you've got algorithms building out these micro-articles from increasingly disparate pieces. Somewhere here there's a line between a preview of a link and republication.
That has come to my attention more recently when looking for short answers to small medical questions surrounding a condition—things I'd forgotten, etc. I quickly learned how off-the-mark those can be.
The feature is really clever and works great when it works, but sometimes just because something is formatted as an authoritative answer, it's promoted above actual correct information. It's almost slower than just looking for a qualifying source in those cases.
I’d argue that they also took on additional legal responsibility when they started censoring search results, whether it’s hard censorship like removing firearm-related products from shopping results or soft censorship like de-ranking right-biased news sources. If you take the entire internet and selectively filter it down, you have the power to craft any narrative as surely as if you were writing the opinion pieces from scratch yourself. It’s hard to call that anything except “editorial power”.
We forgot that the reason search engines were effectively exempt from libel/copyright etc. is because were just describing the web, they published the map but the content was on all the individual websites.
It's a lot harder to argue that model still applies when you've got algorithms building out these micro-articles from increasingly disparate pieces. Somewhere here there's a line between a preview of a link and republication.