|
|
|
|
|
|
by rexreed
1857 days ago
|
|
|
They key here is that transformer-based "search" isn't actually providing links to the sources of information such as how search works now, but rather synthesizing information as a result of being trained on the corpus of Internet data. In this way, Google gets all the value from Internet properties they don't own without having to push any traffic to those sources. So, they get their cake and eat it too. They create a way to regurgitate information from the vast trove of info on the Internet without ever having to share traffic with those sources by moving traffic from their search engines to those sites, like they do now. They get to sell advertising to those who want to capture eyeballs for search results, without having to share any ad revenue with the content providers that are powering that transformer-based search. Ain't it grand? |
|
|
Now that it's pretty close to actually being here, the grim reality is that anyone who was expecting the status quo to just march on like always is going to get screwed over; and the a new wave of successful businesses will adapt to it and thrive.
It's called 'disruption', and it's a bit disappointing to see people here of all places complaining about it.
Sure, I get it, it's google, and if it was some nippy unicorn doing it people would be more enthusiastic, but ML is hard to do right, and having someone who's actually pushing the boundaries of whats possible is, in my opinion, pretty cool.
BERT made a huge contribution, and if this eventually flows out to everyone else to use, that's great news.
...and, if google stops sending traffic to some websites, well, too bad. We'll adapt; so will others.
The ones that can't will disappear.