I'd love for all this to be true, but Google doesn't need publishers.
You know we're already approaching the limit of training LLMs with textual data? There isn't really much more they can consume. The next stage is video so they can actually learn from movement/seeing.
You say that the "AI fails", but it doesn't work like that sadly.. AI doesn't need constant new content from publishers. Once an LLM is trained, it's trained. It can't really learn more from text actually. At this stage it needs video, to see.
Also, what most people don't understand is, LLMs aren't suppose to contain facts/information. The purpose of training them is to give them an understanding of the world. Not specifically knowledge.
Ie, they learn what a cat is, what it's related to, what it does, why it does it.. It's more about it understanding the concepts of language and the world.
It's not meant to be filled with facts like "Who was the 34th President of the US".
For that, you actually use memory, which right now for LLMs is in the form of a vector database. If you can't use the LLM its self for the ultimate source of truth. It's the reasoning engine. You want it to be able to understand and reason, then you pull stuff from memory depending on what it needs to do.
It's like you, if you want to do something, you can google it, read a page, and do it.(Although you too do have memory, although we don't really know much about how memory works in our brains)
The other big thing that you're missing is this :-
Think about how most web pages are created.
Some guy hires a writer to write. Write googles, reads and writes.
The vast majority of web pages are just regurgitated knowledge.
VERY few are direct experience based, and even if they are, where does that experience/knowledge come from? There isn't that much new knowledge being created on earth. Most of it is just existing knowledge.
Really, almost all web pages are junk. A better source are books, which are separate from the web. Most experts will create books, not web pages.
These days there are more sources of knowledge from video though. Many experts will make videos, but again, video is separate from the informational web.
But if you go further, what happens when LLMs have consumed all the video too? THEY are then more expert than humans. They already are in many areas.
What kind of information do you foresee an AI not being able to write about without humans?
If it has access to all textual/video/image knowledge in existence, in every language. What else does it need?
It can watch ALL of youtube. In every language. See every image ever taken and stored online.
Does it need humans to make new knowledge? gpt4 is already smarter in many ways than most humans. I gave it Einstein's Riddle, and it solved it. 2% of humans can solve it. gpt3 didn't, but gpt4 did.
There's also new techniques to make it even smarter like chain of thought. We do this, you just don't realize. When you add in chain of thought, they become incredibly smart, and we're only at start.
So you have something approaching a level of being smarter than humans, 1000's of times faster, with access to all video, images and text in digital form in every language. The ability to write, make videos and make images.
Why does it need Bob from bhw hiring writers on Upwork for his latest affiliate project?
Why expert is going to be able to produce more unique, better content than it?
In summary, AI *already* has all the information. It doesn't need us anymore.
And the legislation part.. Come on
Who controls the world? The little guy? The law is created by those who control the world to protect themselves and keep the world from going into chaos. It's not there to prevent them from achieving their ultimate goals or gaining more power.
You aren't going to stop AI with a lawsuit. It will not work.
Humans ALREADY learn from other people's books/work. Our knowledge is not unique.
An AI is only learning. You can't prevent something from learning and claim copyright. That's like me reading your math book, then making money with math, and you claim I've infringed on your copyright.
It would never work, but most of all it won't work because why would the people in power allow it? vanguard and black rock own most companies. The world is owned by very few people. The people behind the scenes MAKE the laws. Any illusions of fairness for the people are only there for show. No one cares about publishers.
This might seem doom and gloom, but, ignorance of impending change won't save anyone from it. Taking steps to secure the future equipped with knowledge will.
