@shakenvac's banner p

shakenvac


				

				

				
2 followers   follows 0 users  
joined 2022 September 11 00:27:02 UTC

				

User ID: 1120

shakenvac


				
				
				

				
2 followers   follows 0 users   joined 2022 September 11 00:27:02 UTC

					

No bio...


					

User ID: 1120

If you get a surgery done the person operating on you might be on hour 28 and gotten 4 hours of sleep the night before that long ass shift.

I'm from the UK, where a typical long shift for a doctor is 13 hours, so I cant really tell if this is an exaggeration.

But if this is true, holy shit. That is absolutely outrageous. how can you with a straight face protest that doctors are so desperately committed patient wellbeing, while accepting a 28 hour long surgery shift? There's no other way to describe it - that's dangerous. You, above all, should know what the science tells us about decreasing performance with fatigue. Any airline pilot that accepted a shift even close to that would lose their license.

This is a really interesting perspective, but I admit I have a hard time vibing with it. I tried to get into art appreciation when I was younger. Went to the national galleries and the Tate modern, hemmed and hawed at paintings and modern art pieces. This was the top 1% of the top 1% of art, and yet I was disappointed that there was usually very little explanatory notes to go along with the piece. Often when I did find some guide to the 'canon' meaning of the art it was usually perfunctory and not terribly interesting. Usually I preferred my own interpretation to the one I was apparently supposed to draw from the piece. I fully admit this was probably a 'me' problem. Perhaps art appreciation is a deliberately clutivated skill and I simply wasn't able to develop it

All this to say that I'm a 'meaning is in the eye of the beholder' kinda guy when it comes to art. If I draw something meaningful from a piece, I'm not sure it matters if it wasn't the meaning the creator intended, or even if the creator intended no meaning at all.

Besides, what proportion of art that a person consumes on a daily basis actually has layers of meaning deliberately packed into it, let alone deep or philosophical meaning? 1%? Less?

I think that human, natural language definitions of 'stealing', 'plaigiarism', 'copying' etc are not totally fluid. These are words with specific meanings. If someone wants to argue that AI-art is bad on consequentialist grounds then sure, crack on. But 'stealing' is not a catch all term for 'bad'

Whether or not AI-art is bad, I maintain it is not theft.

If your view is that we need to redefine what 'stealing' is in order to specifically encompass what AI does then yes, you can make the argument that AI art is stealing, but if you do that you can make the argument that literally anything is stealing, including things that blatantly aren't stealing.

AI training is novel, but I don't at all agree that it is so novel that it cannot possibly be placed into the existing IP framework. In fact I think it fits reasonably comfortably. I do not believe there is anything that AI training and AI generation does that could be reasonably interpreted to violate any part of IP law, nor the principles upon which IP law is based. You cannot IP protect a style, genre, composition, or concept. You cannot prevent people using a protected work as an inspiration or framework for another work. You cannot prevent people from using techniques, knowledge, or information gleaned from copyrighted work to create another original work. You cannot prevent an individual or company from examining your protected work. You cannot induce a model to reproduce any copyrighted work, nor reverse engineer any from the model itself. Indeed, carveouts in IP law like 'fair use' - which most people who decry AI art would defend passionately - gives far more leeway to individuals than would be required to justify anything generated by an AI.

I wouldn’t want to accuse everyone who is down on AI art as being insincere or a dirty rotten motivated-reasoner -many people freely admit their concern is mainly for the livelihood of artists-, but I have seen these discussions play out many times on many different forums. I have rarely seen the ‘AI-art is stealing’ argument withstand even the barest scrutiny. It is often pushed by people who clearly do not understand how these models work, while aggressively accusing their opponents of not understanding how they work. As @Amadan pointed out in his far-better-than-mine post, when faced with the hypothetical of an ethically trained AI, people do not declare their issues are resolved, which indicates that the core of the disagreement is elsewhere. It smacks of post-hoc reasoning.

Which is evidence that the objections are principled rather than merely opportunistic.

I think the actual root of the objections are sympathetic. Artists are high status in online communities. People see a threat to them, empathise, and develop the core feeling of ‘AI-art bad’. From there we are into arguments-as-soldiers territory. Everyone knows that stealing is bad, so if you can associate AI art with stealing, even if the association makes little sense, then that’s a win.

Yes, the blowback against AI art seems to me a little insincere.

Ostensibly, it's about the AI 'stealing' public art to train itself. (I agree with you that this argument is nonsense)

More realistically, it's people disliking the idea of robots putting artists out of work.

Cynically, it's artists being sore that their highly developed skills can suddenly be near-replicated by a computer in 15 seconds.

Many times over the past few centuries, skilled workers have found themselves driven into obsolescence by technology. Very few of them succeeded in holding back the tide for long. If I were a digital artist, I would urgently be either swapping to a physical medium, or figuring out how I could integrate AI into my workflow.

Is this the chutzpah defence? "We made so many demonstrably false accusations of election rigging that we have no credibility left, which means now the other side can do whatever it wants!"

The boy who cried wolf is like, parables 101.

Who is going to buy an insurance policy where the payout is only twice the premium? Just save the premium.

The fundamental difference between insurance an gambling is that gambling is predicated on the idea that people want to win, wheras insurance is predicated on the idea that people don't want to 'win' (i.e. have to claim on their policy), and this shapes the two industries into fundamentally different things.

That said, I've heard of people using gambling as an emotional hedge, e.g. 'I'll bet on the other team, so if my team loses at least I'll win $100'

Trump has learned nothing, it seems, from past debate performances.

He's almost an octogenarian. even if he did had the attitude and humility required to even want to get better at something, who he is as a person - and what he is capable of - has completely solidified by now. Trump, like basically everyone his age, is nearly incapable of personal growth or learning lessons.

A good point. When you put it that way it’s similar to the lament “I wish I’d worked harder at school” which always sounded to me like “now that it’s reaping time I wish I’d done more sowing”, privileging the wants of your current self over those of your past self.

Very nice, thank you sir!

I wish there were a book called 'Things That 90% of Economists Agree On'. I imagine it would be a pretty thin book, the only thing I know of that would definitely be in it is rent control. But it would at least give a baseline of policies that are so clearly bad that no serious administration should consider it.

Part of the problem is that the profession has lost a lot of credibility. The public has - belatedly - realised that you can always find some economist or other to wheel out in favour of/against almost any policy under the Sun. When that rare issue comes along that unifies the profession it isn't always legible to the public, who have gotten used to ignoring them.

There was an interesting argument I heard that, because the fall of Constantinople (1453) and the colonisation of the New World (1493, Hispaniola) are near contemporaneous, anyone who thinks that the colonisation/conquest of the Americas is fundamentally illegitimate is basically arguing that 'historical conquest' became 'illegitimate landgrab' sometime in that very specific 40 year window.

The simple counter-argument would be that 'stolen' land remains 'stolen' no matter how much time passes, wheras 'gifted' land (or rather, 'gifted' citizenship) is fair and legitimate.