I'm told some of the recent Qwen models were trained on something like 140TB of data or so. I can put some of the Qwen models on a slightly large USB stick (~32G) . No way that contains the whole source text. No form of compression survives that. It's an old argument, it's still the simplest argument-by-numbers.
The intuition this counters is "it's drawing from a big database" or "it memorized everything". It didn't. It still can recall passages at times. But then again, so can I for my favorite books and movies.
And no I'm not saying it's human, to stop the next big jump.
LLMs are LLMs. They have their own properties. Some of which happen to coincide with human properties. Just like a robot arm has some properties in common with humans.
Similar requirements, similar constraints; you're going to have to expect similar behaviors and failure modes at least some of the time. But I haven't heard anyone claim a robot arm can edit a python program, nor have I heard of an LLM picking 100 widgets per minute.
Given the specific of the "derived works" being produced, do you feel obliged to pay royalties every time you cite or use some knowledge acquired from a book during your daily life?
Well no, but I'm a human and not a commercial computer program so it's not particularly relevant. I'm would not be making loads of money making those citations either but that's only secondary.
> derived works
It's not evident that any works are being produced in the legal sense. Since LLM outputs are not considered copyrightable it's probably closer to using a search engine.
The point is that you can buy a book for a few bucks, read it, and have its insights completely available to you. But people are demanding that LLM companies pay far more to publishers for the same copies
You can also buy a book produce a large number of copies after jumbling around some chapters sell it to other people. That's illegal to do if you are person but not if you an AI company.
I suppose an argument might be made for fair use if they released their model weights publicly without financially profiting from it.
Doing that would be considered deeply immoral and sociopathic in any sane culture. Do you really want to live in a society that finds such attitudes acceptable?
Instead of blindly jumping on a manipulated outrage bandwagon, people would do well to maybe read some of those old books, not even the rare ones - they tend to contain plenty of parables and stories explaining basics of morality and civilized conduct. We used to teach that to kids at homes and in primary education...
I've seen many people sell products and courses with price parity, meaning someone in the US pays one price and someone in South Africa pays less.
Me, as a software engineer, will do research on a company to determine how much they can afford to pay me before giving them my salary range.
And almost every SAAS company charges people differently depending on their needs, even if the underlying software is identical between customers.
So no, I don't think it's inherently immoral to charge different prices to people depending on their ability to pay or use cases. It can be, but it's not a defacto standard.
Money is relative, though. I guess the question is if it's acceptable to charge a tiny symbolic fee for that cup of water if the other person's pockets are visibly overflowing with cash. If yes then it's more than reasonably to charge Musk a few million.
Turning it the other way around is it deeply immoral and sociopathic for someone to hoard massive amounts of money/resources if there are people dying or suffering around them (even if not on the spot but e.g. due to poor access to healthcare)?
Given his approximate Trillionaire status, this is just what every vending machine doing income-based algorithmic pricing should charge him for EVERY bottle of water.
Well the argument is that they should pay for the right to produce derivative works not for the physical copy.