> FAANG isn't going to be hiring at anywhere near the same rate they were before, because they won't need to.
They already didn't need to. The reason they hire as much as they do is to keep their lead, essentially denying potential competitors of growth potential by taking the best engineers and trying every crazy idea before anyone else. Or so I think.
And even though they overhire, it doesn't mean they can do huge layoffs without consequences, as the structure is now built around a massive workforce.
Jubijub 19 hours ago [-]
It’s a dark take, but I mostly agree with it.
The ad model, with all its flaws, is a flexible model, and it has funded the web for a while.
The part that will break is that a lot of sites will have 0 incentives to continue to publish, in the face of 0 revenue and 0 credit. That will degrade the quality/ relevance of LLMs.
I also think that “guaranteed produced by humans “ will have value.
blacksmith_tb 13 hours ago [-]
> "The internet is going to return to more of its original roots, which are niche fan websites you largely find through social media or word of mouth"
Ok, that sounds possible, and not so bad...
> "Very few of them are going to survive... it costs a lot of money to pay people to write high quality content"
That may also be true, but doesn't really jibe with the image of the early 'net?
weregiraffe 11 hours ago [-]
Early net was tiny.
hdjdbdirbrbtv 10 hours ago [-]
But also unfathomably large because of the speed at which you could consume it was measured in kbps....
palata 19 hours ago [-]
In one sentence, this is what I have been saying about LLMs since they got impressive:
"I don't know what good it can make, but this thing clearly has the potential to break the Internet".
xg15 21 hours ago [-]
Tell me again how LLMs were going to make everything better?
raxxorraxor 8 hours ago [-]
I use them for coding and other AI for image/video generation. It is awesome.
But I have yet to see a convincing consumer product leveraging LLM AI. In our company it prefilters support mails. Tries to connect customer requests to customer information in our databases. Maybe connect device information if a mail contains serial numberes. Some convenient stuff here and there to provide more context, but it certainly doesn't have any critical roles.
And developing LLM services takes a huge amount of time, comparable to other software projects.
palata 1 days ago [-]
> Paid services like Kogi will be
Typo: it's called "Kagi"
shmeeed 19 hours ago [-]
This is a chilling outlook, but it looks very plausible.
The process has already started, the building blocks are in place. See the recent rise in public complaints about intense scraper activity. Zero-Click has become all but inescapable, and is going to capture an ever-growing share of searches.
Still, any paradigm shift also gives room for hope. There will be pitfalls for Moloch, they might just trip over their own ambitions. And maybe there will be opportunity for organic growth to take root in the cracks of their foundations.
homeonthemtn 3 hours ago [-]
Yes and there will still be a "free" version of content somewhere because not everyone will pay those prices (because they can't afford them or didn't know the value of a premium service).
So since there will be a demand, someone will produce "free resources" which will be the equivalent of a sewage outlet running into the sea.
Absolute garbage of an AI ouroborus just churning in on itself to aggregate and produce something for whatever this free channel of content ends up being.
theodorewiles 13 hours ago [-]
I think end state is LLM-facilitated micropayments. One vast clearinghouse / marketplace of human-generated up to date content. Contributors get paid based on whether LLMs called their content via some kind of RAG. Maybe there are multiple aggregators / publishers.
dzonga 21 hours ago [-]
for now - we can identify AI slop.
in the future - yeah interaction will be at a premium and that includes all textual content - since you can't trust whether it was generated by AI or not. which means books / content pre-AI gonna more valuable.
Later on things will need to be human curated.
Havoc 15 hours ago [-]
If you genuinely can’t tell the difference anymore is there any motivation to pay a premium?
notnullorvoid 14 hours ago [-]
It depends, can you not tell the difference because you don't know better, or you can't tell the difference because the quality and accuracy is actually good.
Only to keep human authors employed -- Once an art is lost, it's lost, and the llm cartel can only continue to operate at a loss for so long before the piper needs paid
hiccuphippo 11 hours ago [-]
Human authors will not write but direct the AIs on what to write. Then they'll syndicate that to AI companies so they can keep their AI fresh. Like an ouroboros eating its own tail.
8 hours ago [-]
Rendered at 16:01:54 GMT+0000 (UTC) with Wasmer Edge.
They already didn't need to. The reason they hire as much as they do is to keep their lead, essentially denying potential competitors of growth potential by taking the best engineers and trying every crazy idea before anyone else. Or so I think.
And even though they overhire, it doesn't mean they can do huge layoffs without consequences, as the structure is now built around a massive workforce.
The part that will break is that a lot of sites will have 0 incentives to continue to publish, in the face of 0 revenue and 0 credit. That will degrade the quality/ relevance of LLMs. I also think that “guaranteed produced by humans “ will have value.
Ok, that sounds possible, and not so bad...
> "Very few of them are going to survive... it costs a lot of money to pay people to write high quality content"
That may also be true, but doesn't really jibe with the image of the early 'net?
"I don't know what good it can make, but this thing clearly has the potential to break the Internet".
But I have yet to see a convincing consumer product leveraging LLM AI. In our company it prefilters support mails. Tries to connect customer requests to customer information in our databases. Maybe connect device information if a mail contains serial numberes. Some convenient stuff here and there to provide more context, but it certainly doesn't have any critical roles.
And developing LLM services takes a huge amount of time, comparable to other software projects.
Typo: it's called "Kagi"
The process has already started, the building blocks are in place. See the recent rise in public complaints about intense scraper activity. Zero-Click has become all but inescapable, and is going to capture an ever-growing share of searches.
Still, any paradigm shift also gives room for hope. There will be pitfalls for Moloch, they might just trip over their own ambitions. And maybe there will be opportunity for organic growth to take root in the cracks of their foundations.
So since there will be a demand, someone will produce "free resources" which will be the equivalent of a sewage outlet running into the sea.
Absolute garbage of an AI ouroborus just churning in on itself to aggregate and produce something for whatever this free channel of content ends up being.
in the future - yeah interaction will be at a premium and that includes all textual content - since you can't trust whether it was generated by AI or not. which means books / content pre-AI gonna more valuable.
Later on things will need to be human curated.