ChatGPT For World Building?

[Demesnedenoir]

Spellchecking and "suggestions" are the infant versions of AI in writing/language in the same way as some of the features in Photoshop over the years were the start of visual AI as we know it now. There is no "intelligence". So, what's AI and what isn't is more of a brand name than reality, open to interpretation.

 

[Finchbearer]

I’m not a computer no-it-all, I’m just fairly sure that AI tech is different from regular computing. Something like a spellcheck feature on any word processor is a fixed input of programming - someone programmed that function to perform one task only from a fixed source, in that case is the dictionary. I don’t know how AI is done in technical terms, but it seems more like it programming so that the computer is making choices on its own rather than from a ‘script’.

 

[Devor]

^ Right, it's only considered AI if the system has the power to learn from new data and adjust its own output based on that information. A typical spellcheck as most people think of them wouldn't be an AI. It's drawing on a style book and dictionary and is always going to give you the same response. In fact, in Word, you can (used to? it's been a while) find a list of the autocorrections it makes (like teh = the) and click them off; it didn't look at the data and come up with them itself, instead a person picked them out.

Other things might look a bit like AI, but they're choosing randomly from a list. An AI would be able to decide what's on that list, or at least the percentages assigned to each option, and then update them in real time.

Nowadays they are, like everything else, starting to use AI. I haven't been using them lately so I'm not quite sure on specifics.

 

[pmmg]

I am not sure this is worth hashing out, cause as AI moves forward, spellcheck is becoming more just a part of the overall AI package.

If not already, then sometime soon, the Spell Check Engine will be one in the same with the AI engine, and to say one will be the same as saying the other.

Already, the online grammar companies are advertising their new AI features as a way of saying choose us, we have better grammar checkers, we have AI.

But...technically, unless the computer is making self aware choices, its not truly AI. Its just something doing what its code told it to do.

The question for all of us is not 'is it a spell checker or is it AI', but what is your true contribution if you are just regurgitating what comes from AI anyway? The less of you, and the more of it, and the further away you are from even mattering. I want you to be special, and not just the next silly person saying look what I did with AI? Sorry...but you were not needed.

 

[skip.knox]

I still find Douglas Hofstadter's take on the matter of artificial intelligence to be useful. What it is in theory is more or less irrelevant, he argued (I hope everyone is ok with me putting words into the mouth of a winner of the Pulitzer Prize). What matters, the tipping point, will be when humans call it intelligent. When we start treating the machine, the algorithm, the device, the whatever, as if it were intelligent. To put it another way (sorry Doug), humans are never going to agree on what we mean by "intelligent" but there will develop some sort of common usage.

It is, imo, too soon for that. It's still developing. I am brash enough to venture an addition to Hofstadter here, and say it will be intelligence when it is recognized in law. We're seeing the first stirrings in that direction, but some day it will come to debates over things like citizenship and rights and terminations and suchlike.

But right now, despite the technical marvels, it's mostly ad copy. Just because we're on the way, doesn't mean we've arrived.

 

[Fyri]

^ Right, it's only considered AI if the system has the power to learn from new data and adjust its own output based on that information.

I've been seeing things that suggest that mostly no one really agrees on a definition for AI today, but this is a nice distinction.

 

[Devor]

I've been seeing things that suggest that mostly no one really agrees on a definition for AI today, but this is a nice distinction.

You get a lot of arm chair commenters, but it can also get a little muddy. For instance, when I lose at Super Smash Bros to a level 3 AI (don't judge me), there's no AI on the game cartridge. It's more like a snapshot of an AI tweaked to a specific play level; it's locked in. But the program was definitely made by an AI model to simulate a human player, so nobody's going to bat a-eye* if you call it one.



*I think I need to go to confession for that one.

 

[CupofJoe]

I've been seeing things that suggest that mostly no one really agrees on a definition for AI today, but this is a nice distinction.

The best definition of AI I've heard is "a bad choice of words in the 1950s".
It is all computational statistics.

 

[_Michael_]

If there is a specific AI function in newer versions then that specific feature is clearly artificial intelligence but psd is a toolbox - you can’t just input a prompt and it does the work for you, and I don’t think that’s ambiguous. I refuse to pay for the creative cloud so I’m running probably a vintage model, but it is my preference for how much I actually utilise it lol

How then are you able to say this with a straight face while turning around and telling others that AI tools like ChatGPT aren't just tools the same? ChatGPT is just a fancy tool with shiny packaging. Still requires a lot of input and fiddling with to produce anything useful...just like using the PSD toolbox.

My point was always that ChatGPT and other AI generative tools are just that--tools. Still requires someone to actually use them, and still requires a human to refine and combine and edit and tinker with, just like you have to do in PS or CorelDraw. That the tool can do more things is irrelevant; this is like saying that a Swiss Army knife is no longer a knife because it has other functions in addition to cutting.

Unless and until the US intelligence agencies decide to let us play with their AGI tools (probably never because that would remove their ability to surveil every single person on the planet), the current AI in the public sector is a DARPA-forged Frankensteinian creation that uses fuzzy logic and is deliberately made idiotic to keep it from being too useful. Prove it? Sure. Remember how Siri, when she first came out, was able to have conversations, was pretty witty and funny, and would answer non-critical queries with humor? Now, she can barely manage an address book. That is a definite downgrade from where she was originally, and lightyears behind what actually came out of DARPA's research.

Only true AGI could be called a danger (and it is, and it's already in the wild a la PROMIS and the CIA's malfeasance) or a threat to artists since that is an actual sentience behind the software capable of learning, adapting and rewriting itself without human input. You can lol all you like about government conspiracy theories, but there's more than enough evidence to prove this is exactly what is happening. Moreover, it's highly likely that the AI distros in the public sector right now are actually secret moles designed to simply gather information on all users and user habits to flesh out and refine the PROMIS landscape to show live-time who is where and doing what at any given moment of the day.

The simplistic generative AI suites floating around today are but a shadow of what the MIC actually has, and until the AGI programs come out, calling these AI programs anything other than overproduced tools is giving them too much credit. The human operator (or artist, in casse of writers and designers) must still parse what the AI spits out and work it into their project, or muck about until they get exactly what they're looking for. Arguments about derivation are irrelevant; the onus is on the human operator to recognize and remove any trademark or copyright violating material, just as it would be for using any other software tool. CorelDraw artists aren't going to knowingly duplicate Microsoft's logo, for instance, in the course of their work.

The tool (AI) spits out the seed of an idea. The human then takes that and bashes it and squeezes it and shapes it to fit their vision. If the artwork or literature is derivative, then it is because the artist himself or herself is being lazy and, like what happens in academia, when we find those artists, we avoid them, they fall out of the market with their terrible, derivative ideas, and new and better ones replace them. A digital freemarket in other words. We know the model works because Reddit is still around.

^ Right, it's only considered AI if the system has the power to learn from new data and adjust its own output based on that information. A typical spellcheck as most people think of them wouldn't be an AI. It's drawing on a style book and dictionary and is always going to give you the same response. In fact, in Word, you can (used to? it's been a while) find a list of the autocorrections it makes (like teh = the) and click them off; it didn't look at the data and come up with them itself, instead a person picked them out.

Other things might look a bit like AI, but they're choosing randomly from a list. An AI would be able to decide what's on that list, or at least the percentages assigned to each option, and then update them in real time.

Nowadays they are, like everything else, starting to use AI. I haven't been using them lately so I'm not quite sure on specifics.

So what's the difference between spellcheck offering you suggestions in a list, and ChatGPT doing the same with plot hooks or artwork? You generate some material, you pick out what you like, work it over and refine it, and presto--it's no longer the product of AI derivation.

 

[Finchbearer]

Wut?

I’m saying it’s a different type of computing.

 

[Finchbearer]

I’ll just leave this article here that explains far better what I’m trying to explain. Difference between AI and regular computing.

 

[_Michael_]

I get that, but there are too many similarities to normal computing for it to be anything other than a fancy tool box. Take Bethesda's Garden of Eden Creation Kit (the GECK). It has AI tools built into it (search queries, script debugging, script compiling, error sensing, etc). Does that make it derivative simply because users can choose to use it to open BSA files (prohibited by Bethesda) and copy them? Of course not. Just like any other tool that can be used for good or bad, it's the user who makes it or breaks it.

When you discuss derivation, that's a red herring because all of human culture is derivative. That AI does the copying much more clumsily than human artists is simply a product of the crude and unrefined nature of the AI software suites in their present level of development. The AGI tools like LaMDA are, in fact, sentient, and those programs could rightfully be accused of plagiarism or derivative work because they have the ability to learn right and wrong, but their thinking is completely alien to that of humans. PROMIS, for instance, is everything you fear AI could be, and worse, and it's out in the wild. It's already monitoring us.

AI generative programs are just fancy toolboxes incapable of sentient comprehension of what they're producing, just as the paintbrushes and palette are not capable of sentient comprehension of the artist's work. The only difference between the two is one is digital. That's it. Someone who uses them to produce copyright and trademark violative material will be caught and then publicly tarred and feathered. The artist who uses AI instead to produce new idea seeds and bounce ideas off of, using it like a personacom in Chobits, will produce better material because of access to better tools (AI).

 

[Finchbearer]

You’re talking about the moral question of ‘is AI derivative’ - I’ll give you an example; when I was looking for Arthur Rackham’s artwork for another thread I started on here, I found someone Midjourney ‘portfolio’ with a load of Arthur Rackham prompts - now that’s not derivative, that’s just copying, but what, just because a computer has done it then it’s okay? I would say no.

 

[_Michael_]

You’re talking about the moral question of ‘is AI derivative’ - I’ll give you an example; when I was looking for Arthur Rackham’s artwork for another thread I started on here, I found someone Midjourney ‘portfolio’ with a load of Arthur Rackham prompts - now that’s not derivative, that’s just copying, but what, just because a computer has done it then it’s okay? I would say no.

I've already answered this to some extent.

That AI does the copying much more clumsily than human artists is simply a product of the crude and unrefined nature of the AI software suites in their present level of development. The AGI tools like LaMDA are, in fact, sentient, and those programs could rightfully be accused of plagiarism or derivative work because they have the ability to learn right and wrong, but their thinking is completely alien to that of humans.

And again, you're trying to impart some willful plagiarism to a tool that is not a true AI. If there's any derivation, it is the fault of the human programmers of that tool, no different than if CorelDraw's functionality were broken by an update, and the fault of the users, who are the "will" behind the AI. The "AI" is a misnomer. It is a fuzzy-logic generator. That's it. All of them are derived from research that came directly from DARPA and Rockwell Aerospace. I have the receipts to prove it.

This, in turn, also again boils down to the responsibility to respect trademarks and copyrights by the user of the tool (otherwise known as "being an adult,") not the tool itself, which I have also answered before.

Arguments about derivation are irrelevant; the onus is on the human operator to recognize and remove any trademark or copyright violating material, just as it would be for using any other software tool...

The tool (AI) spits out the seed of an idea. The human then takes that and bashes it and squeezes it and shapes it to fit their vision. If the artwork or literature is derivative, then it is because the artist himself or herself is being lazy and, like what happens in academia, when we find those artists, we avoid them, they fall out of the market with their terrible, derivative ideas, and new and better ones replace them. A digital freemarket in other words. We know the model works because Reddit is still around.

Lastly, the idea that the artist's technique is somehow trademarked is ridiculous. If that was true, individual schools of art would never have developed. Impressionists paint in the style of Claude Monet, Aflred Sisley, and Pierre-Auguste Renoir, specifically copying their techniques. Same goes for classicists, realists, cubists and others. We derive and improve. So if someone's work is copying the techniques of Renoir, that's okay but copying someone like Michael Monte Moore is not, why? Because one is dead and the other is not?

"AI" generators are simply a tool that has no will or sentience in and of itself because it's not an AGI. So, your problem in the above scenario wasn't with the AI generator, it was with the fact that what you perceive as someone's prompt ownership because Arthur Rackham is still alive. What if had been, say, filled with Boticelli prompts instead? The human trend is to continue to divide and stratify, which means it's likely that as centuries pass, there will be greater and greater categorization of art schools, or schools will become meaningless because techniques will instead solely be associated with an artist's name for digital purposes.

 

[Finchbearer]

Copying is copying. If someone were to take prints from their Midjourney account in the style of [insert artist] then that would be copyright theft. Just the same as someone in real life making a copy of a famous painting and trying to make a profit from it. It’s happened many times before and will continue to happen. With AI what’s murky about it is that because it’s a third party operator somehow it gets around copyright theft. Hopefully not for long.

 

[Finchbearer]

Arthur Rackham is long since dead. He spent an entire career as an artist illustrator. Boil a lifetime of work into an AI prompt I think is a pretty disgusting lack of respect if you ask me, oh and copyright theft.

 

[_Michael_]

Copying is copying. If someone were to take prints from their Midjourney account in the style of [insert artist] then that would be copyright theft. Just the same as someone in real life making a copy of a famous painting and trying to make a profit from it. It’s happened many times before and will continue to happen. With AI what’s murky about it is that because it’s a third party operator somehow it gets around copyright theft. Hopefully not for long.

There are already existing laws to protect artists and software designers, as well as punish them for plagiarism. We're not talking about copying exact images, we're talking about techniques and styles. Someone outright just copying portions of someone's art should be prosecuted.

Arthur Rackham is long since dead. He spent an entire career as an artist illustrator. Boil a lifetime of work into an AI prompt I think is a pretty disgusting lack of respect if you ask me, oh and copyright theft.

Then you also agree that Photoshop should never be used since they have an AI filter that copies pointilist and impressionist styles of artwork, right? Since Adobe basically boiled down lifetimes of work to create the palette style.

What it really sounds like is anger over the length of time it takes to create a work. If an AI generator artist or author, as in the case of ChatGPT, can generate half-decent subject matter that can be further tweaked in half the time it takes a traditional artist with a tenth of the effort, suddenly, that's bad and the artist is "lazy" and producing derivative work, simply because they're not spending as much time as a traditional artist.

I'm reminded of another radical that got accused of that very same thing when he produced quality works in 45 minute slots on PBS in oil paints because he didn't want to duplicate classical painting techniques and wanted something accessible by everyone. "Ermergerd! He's terkin' er jerbs!" the "classical" critics cried. And you know what? World moved on, and now, people that want to dedicate a few months to themselves can spit out amazing landscapes using this man's technique that takes a tenth of the time "classical" styles required to layer a painting.

Exact same scenario. And yet, the man's paintings literally hang in the Smithsonian now and he's considered a national treasure.

"AI" generators are just the digital equivalent of this, only, the digital Bob Ross has not really emerged, yet.

 

[Finchbearer]

You’re just wrong on every point. Simple as.

 

[skip.knox]

These points are wrong, for reasons x, y, z.

Saying "you're wrong" is an ad hominem argument and violates the rules of the forum. There's no benefit and much potential harm down that road. I also have to point out that a debate on the originality or even existence of AI does not speak to the original post. Let's let this one lie.

 

[Finchbearer]

~ sorry ~ I was tired of the word soup by that point.