martin-t

36 ms·

511.
by martin-t 1y ago
> I don’t think copyright works that way. Maybe but it should - see sibling comment. Statistics as in taking a large input and processing it into much fewer values which describe the input in some relevant ways (and allow reproducing it)
512.
by martin-t 1y ago
> fwiw, I mostly agree with you (ai training stinks of some kind of infringement), but legal precedent is not favouring copyright holders at least for now. I know. I am describing how it should be. Copyright was designed in a time when c
513.
by martin-t 1y ago
The issue is calling it "AI" in the first place. It's not, it's a statistical model of existing text. There's no "genAI", there is "genMS" - generative machine statistics. When you look at it tha
514.
by martin-t 1y ago
> There aren't any insults in their posts. Intellectual dishonesty is a form of insult. As is, for example, "laughably false". If you think I am wrong, you can explain why. But you choose to insult me because you enjoy it.
515.
by martin-t 1y ago
You brought capital-intensive business into it and it's a fair example which needs more elaboration but now you're acting like that's the only example that matters. There are plenty of businesses, not just "tiny" on
516.
by martin-t 1y ago
> It misses the forest for the trees. Yes, this is something I can agree with - many people are aware of societal issues in the small (abusive people they interact with personally, specific instances of injustice which affect them person
517.
by martin-t 1y ago
> You don't even know what we're discussing Stop insulting me. > I quoted in my op comment. I considered you meant this but dismissed it because what you said clearly does not follow from it. A recipe takes experimentation -
518.
by martin-t 1y ago
And is that right? Is that how it should be? I will partake in the taking because ultimately the world is PvP now and doing otherwise would disadvantage myself against those who would. But I will not support such a system. At least your exa
519.
by martin-t 1y ago
That's incredibly dismissive
520.
by martin-t 1y ago
What quoted text? > The stance is incoherent. Mine? Explain how. Yours? Certainly: > your glib and punchy response: its more important that people are rewarded for their ramen recipes than it is for the masses to have access to the ge
521.
by martin-t 1y ago
Yes, despite all the hate GDPR gets from people who have to implement it and from companies whose business model is parasitic, it does seem to go in the right direction. However, I doubt it can be extended to training statistical models. LL
522.
by martin-t 1y ago
Yes, it's absolutely essential that people are rewarded for intellectual work, otherwise they'll stop doing it. "The masses" have absolutely no right to demand I hand them what I produce, whether physical or intellectual
523.
by martin-t 1y ago
One law I would like to see if expected durability. Food has an expiry date and ingrediant list. Something similar should accompany all products so consumers can make an educated choice how long it's gonna last and what's gonna br
524.
by martin-t 1y ago
You really did not run into a single argument against A"I" because of plagiarism, copyright infringement, LLM-induced mental illness, destruction of critical thinking skills, academic cheating, abuse of power / surveillance,
525.
by martin-t 1y ago
It's not about the ramen being good or bad. It's about the recipe having artistic, intellectual and monetary value because human time was spent to produce it.
526.
by martin-t 1y ago
And that's just one part of it. Part 2 is all the copyleft code powering the world. Now it can be effortlessly laundered. The freedom to inspect and modify? Gone. Part 3 is what happens if actual AI is created. Rich people (who usually
527.
by martin-t 1y ago
I was gonna make an analogy of stealing someone's screwdriver set when I need to solve a unique problem but this is so much better.
528.
by martin-t 1y ago
Quantity matters. Imagine it took 10^12 hours to produce the training data, 10^6 hours to produce the training algorithm and 10^0 hours to write a bunch of prompts to get the model to generate a useful output. How should the reward be distr
529.
by martin-t 1y ago
In most of those cases, it was because too many people broke the laws, regardless of what companies did. It was too distributed. But to train a model, you need a huge amount of compute, centralized and owned by a large corporation. Cut the
530.
by martin-t 1y ago
I went through exactly this process. Then I discovered (A)GPL and realized that the system makes sense to protect user rights. And as I started making my own money, I started paying instead of pirating, though I sometimes wonder how much of
531.
by martin-t 1y ago
Yep, just like it's illegal to sell your organs or sell yourself into slavery, society needs to recognize that even through the severity of exploiting personal data is much lower, the principle is the same - the power differential betw
532.
by martin-t 1y ago
And this is exactly the problem. It used to be the case that you exchanged money for a good or service. It was a transaction of 1 thing for 1 thing. Now you're exchanging your money AND personal information for goods or services (somet
533.
by martin-t 1y ago
Here's how the law should work. You own the data you produce, both intentionally (writing, making videos) and unintentionally ("metadata", logs). You have to explicitly give others permission to use that data for any purpose
534.
by martin-t 1y ago
This shouldn't be enforced through technology but the law. LLM and other "genAI" (really "generative machine statistics") algorithms just take other people's work, mix it so that any individual training input i
535.
by martin-t 1y ago
The average blogger is a male and the parent commenter is perfectly OK with average opinions, average summaries, etc. so it makes perfect ironic sense that he doesn't care.
536.
by martin-t 1y ago
LLM have been shown to not summarize the actual content of what you give them as input but some statistical mashup of their training data and the input. So they will misrepresent what you in the end, pushing the readers (note not "your
537.
by martin-t 1y ago
It always starts with people doing real positive-sum work and then grifters and parasites come along and ruin it. We could make advertising illegal: https://simone.org/advertising/
538.
by martin-t 1y ago
> Of course they are; if someone's going to get money from VCs they risk their reputation in asking for it. You are unintentionally proving me right. There 2 kinds of people starting businesses which need an investment - those riski
539.
by martin-t 1y ago
> Any business that needs money to start needs people to risk their reputation or their money to get it going Nobody is risking their reputation, that's just being over-dramatic. Money? Sure. There's two kinds of people here: a
540.
by martin-t 1y ago
I can't believe the hypocrisy of a guy with 76029 internet points (that's a big time investment, would be a shame if someone trained an LLM on it) pretending to not understand that people want recognition for what they say, regard

More ›