visarga

40 ms·

811.
by visarga 2y ago
It might be a good idea to try vf-vibe-coding
812.
by visarga 2y ago
How about selection? If I select the good image from 1000 others? Curation is also a contribution to art.
813.
by visarga 2y ago
> you can't copyright something that your AI generated Seems like a loophole, if I generate synthetic data with a model trained on copyrighted works, the synthetic data is copyright free? So I can later train models on it?
814.
by visarga 2y ago
So.. does the conductor of an orchestra get royalty rights? He's just prompting the "actual" musicians.
815.
by visarga 2y ago
Maybe the best idea would be just to scrap copyright alltogether. It just blocks people from collaborating and building on top of each other's work. If everyone demanded royalties, where would Linux be? Wikipedia? scientific research?
816.
by visarga 2y ago
> Things made with AI should be protected by copyright, with the rights held by the user of the tool that generated the image. Like any other digital art. I would agree for carefully crafted outputs where the human had a major contributi
817.
by visarga 2y ago
> no other humans creative input and therefore livelihood was ever involved in the process What would be the creative output of an artist who never saw the creative output of other artists? We think too highly of ourselves, as if creativ
818.
by visarga 2y ago
Estimating the number of words used by humanity: starting from total number of humans who ever lived 110B, multiply that with 80 year * 365 days * 35000 words ~= 1B words/human lifetime, you get 1.1*10^20 words. Now, the training set o
819.
by visarga 2y ago
> LLMs are interest case, but they are nearly flat. They are less specialized than brains, but compensate millions of years of evolutionary adaptation with a larger corpus of text, about the same size as 11,000 people use in their lifeti
820.
by visarga 2y ago
Chomsky has that wrong theory about language innateness. In reality languages that don't fit child brains don't survive. It's the other way around!
821.
by visarga 2y ago
> Clearly false (what words do you use when you plan to throw a ball to a target?) If a monkey can do it, it's not very abstract.
822.
by visarga 2y ago
I can ask a LLM to write a haiku about the loss function of Stable Diffusion. Or I can have it do zero shot translation, between a pair of languages not covered in the training set. Can your "language JPEG" do that? I think "
823.
by visarga 2y ago
LLMs are much easier to port than software. They are just a big blob of numbers and a few math operations.
824.
by visarga 2y ago
The counter perspective is that this is not a book, it's an interactive simulation of that era. The model is trained on everything, this means it acts like a mirror of ourselves. I find it fascinating to explore the mind-space it captu
825.
by visarga 2y ago
Are you implying AI is causing more competition than other creatives, and the billion choices the public has?
826.
by visarga 2y ago
The transformer is a simple and general architecture. Being such a flexible model, it needs to learn "priors" from data, it makes few assumptions on its distribution from the start. The same architecture can predict protein foldin
827.
by visarga 2y ago
You say that, but the reality is that all open models rely heavily on synthetic data generated with ChatGPT. They don't like it, but it happens anyway. You can't really protect a public model from having its outputs exfiltrated. T
828.
by visarga 2y ago
But who should pay? The model developers? Training models is a cost center. And what about open source AI, should we legislate it out of existence? How about the AI providers? they operate on thin margins, and make just cents a million toke
829.
by visarga 2y ago
> Can't believe I'm actually rooting for the copyright cartels in this fight. The same megacorps are suing Internet Archive for their collection of 78rpm records. These guys would rather see art orphaned and die.
830.
by visarga 2y ago
But making copies for yourself, without distributing them, is different than making copies for others. Google is downloading copyrighted content from everywhere online, but they don't redistribute their scraped content. Even web browsi
831.
by visarga 2y ago
I think you are right, the value is in application. If you have a problem to solve with AI, you benefit from AI. If you don't, you don't.. like any software. The AI providers get $20/month, you can get anything out of AI. Use
832.
by visarga 2y ago
The ChatGPT organization of conversations is terrible - you can't even read the full titles - you have to use an infinite scroll list clicking "see more" every time - search is very bad, and not even using semantic search, yo
833.
by visarga 2y ago
Feels like the JavaScript framework wars era all over again. Wondering what will be the prevailing paradigm.
834.
by visarga 2y ago
> it takes a bit to find out they cannot code, and a bit longer to fire them, so you're out $$$$ paying their salaries for nothing Not just that. They block the hiring process. Maybe there was just that one open position on your tea
835.
by visarga 2y ago
You don’t need legal solutions for something that technology can fix. Agents can gather, filter, rank, and display information in a way that works for you. Gatekeepers control both the information and the interface, but agents take that con
836.
by visarga 2y ago
I think the opposite, just like we used to tell people to "go read a book", now we'll tell people to "spend some time with an AI" to get cultured. The more AI time the better for your education.
837.
by visarga 2y ago
> Ads in the AI overview. That's always been the endgame for them here. The fundamental problem of ads is that they sell other things than we need. LLMs won't change that.
838.
by visarga 2y ago
Disappointing not to have any of this in EU.
839.
by visarga 2y ago
> Game programs like AlphaGo and AlphaZero (chess) are all brute force at core - What do you call 2500 years of human game play if not brute force? Cultural evolution took 300K years, quite a lot of resources if you ask me.
840.
by visarga 2y ago
> if you do AI research with the idea that by teaching machines how to do X, we might also be able to gain insight in how people do X, then ever more complex statistical setups will be of limited information At the very least we know con

More ›