[drewprops]

We don't have a catch-all for this topic, which will bridge lots of disciplines and applications. Please relocate this as makes sense.

I wanted to share that I've found an interesting application for Chat GPT; therapy in talking to people who you can't talk with either through death or distance.

What are some applications you're using now?


...

[Kickaha]

Can we agree that these are not AI except in marketingspeak?

They are very nice plagiarism engines, however. Top notch.

[drewprops]

Originally Posted by Kickaha Can we agree that these are not AI except in marketingspeak?

Yes, thank you. I am being lazy, and (still) largely unable to comprehend the mechanisms used to synthesize intelligence. The moment a bot talks back to me in Steve Jobs' voice, I am hoodwinked by this modern Turk.


...

[PB PM]

Originally Posted by Kickaha Can we agree that these are not AI except in marketingspeak? They are very nice plagiarism engines, however. Top notch.

100%. They are not just plagiarism machines, but they are also generating fake stuff too.

There is a story on Ars today about the lawyer who used ChatGTP to do some of his work, problem was the case examples the AI gave him were all fake and now he is in hot water with the judge.

[709]

Originally Posted by Kickaha plagiarism engines

stealing

Originally Posted by drewprops modern Turk

also stealing

[PB PM]

Originally Posted by 709 stealing

That is what plagiarism means.

[Ryan]

stochastic parrot

[indignantritzy]

Originally Posted by PB PM 100%. They are not just plagiarism machines, but they are also generating fake stuff too. There is a story on Ars today about the lawyer who used ChatGTP to do some of his work, problem was the case examples the AI gave him were all fake and now he is in hot water with the judge.

just plagiarism machines but the amount of knowledge they have is greater than that of humans.

[drewprops]

Ever been to Toledo, buster?


...

[kscherer]

We're all very nice around here. Those of us that aren't from "that" place, anyway.

[Kickaha]

Speak for yourself, I'm a jerk.

Saw a comment recently that is so damned on point:

"Everyone notices when a ChatGPT style engine generates completely false output, and call it 'hallucinating', but no one seems to understand that it's also hallucinating when it produces something that is accurate."

It's *all* BS in side the ChatGPT, the only metric of whether it did it right is when it is reviewed by a knowledgable human. When it gets it right, it is just as accidental as when it gets it wrong.

[kscherer]

Yes, you are a jerk!

That said, the only thing I know about all this ChatGPT BS is that politics 2024 is going to be an absolute shit-show!

And I can't wait!

The fake photos and videos that are going to parroted about as truth will be head-snappingly fun; the fake "he-said, she-said" letters to and from concubines is going to be absolutely marvelous; the fake boy-scout-butt-pinching will be a political gas-fest. This is going to be the best and most comedic election cycle in the history of election cycles.

It is going to be so much fun watching politicians tap-dancing on their respective party's platforms that popcorn will run out of stock on all the local shelves.

I. Can't. WAIT! I cannot wait to see what people do with this "smarter than humans" nonsense. Out the other end can come nothing but hilarity best observed from a very long distance.

[PB PM]

I’m looking forward to better and smarter bots to joining the forum, maybe they’ll do more than copy and paste what previous posters said already. Even if what they say is, just a fluke. They can all go to Toledo together later!

[drewprops]

I am listening to a podcast interview with an attorney who is developing an ethical AI tool for tracking the sources used to deliver results from AI generation (in this case, for film scripts).

I fundamentally do not understand (and this is my naïveté and ignorance of the law speaking) how tracking sources is useful or even possible once you move out into second generation and beyond.

Let's say that this proposed ethical AI technology works, and uses blockchain to document that my new AI-generated script combines Bladerunner and Gone With The Wind (for the purposes of compensating those copyright holders).

Then Paul asks his ethical AI to create a script that blends Star Wars with my film.

Per this fellow's proposed system, the copyright holders for my film and any copyright holders that I used to create my film would be compensated in some fashion.

Beyond the question of "how many generations out can copyright holders expect to receive compensation?" there's the question of (in my mind) "Prior Theft".

Because, as any true Star Wars fan knows, as groundbreaking as the universe of Jedis and Sith was when it first debuted, large parts of the story were inspired by (if not liberally stolen from) other sources, like Lensman.

I believe that "ethical AI" in the fashion that this man proposes ignores the entire history of human storytelling - regenerative techniques and all.

Will they use an AI detective to go back and trace the inspirational sources for pre-AI films, to reward the creators who were never compensated for their purloined intellectual property?

I mean, wouldn't that be ethical?

How far back do we go?

...

[Kickaha]

Originally Posted by drewprops How far back do we go? ...

Backdate to limit of copyright law, would be the naive guess.

Of course, that keeps getting extended, so...

[Brad]

Also, just to be clear,

Originally Posted by drewprops I fundamentally do not understand […] "ethical AI"

Nobody actually understands it. It's all smoke, mirrors, and snake oil. "Ethical AI" is just the latest buzzword to try to sway public opinion and fleece naive investors while regulations and legal precedence are sparse.

Very few, if any, useful models have been "ethically" trained on purely public domain content or where all copyright holders have explicitly and with full knowledge opted into its training and derivative uses, and it takes a lot of content (like an unfathomably large amount) to train a useful model.

Originally Posted by drewprops how many generations out can copyright holders expect to receive compensation? […] Will they use an AI detective to go back and trace the inspirational sources for pre-AI films, to reward the creators who were never compensated for their purloined intellectual property?

Zero and no, in that order, I believe I can say with reasonable confidence.

Unless Disney is running the system and, well, they just so happen to find that they deserve compensation.

[Brad]

Also also, I'm long overdue for renaming the thread.

[Ryan]

Originally Posted by Brad Also, just to be clear, Nobody actually understands it. It's all smoke, mirrors, and snake oil. "Ethical AI" is just the latest buzzword to try to sway public opinion and fleece naive investors while regulations and legal precedence are sparse. Very few, if any, useful models have been "ethically" trained on purely public domain content or where all copyright holders have explicitly and with full knowledge opted into its training and derivative uses, and it takes a lot of content (like an unfathomably large amount) to train a useful model. Zero and no, in that order, I believe I can say with reasonable confidence. Unless Disney is running the system and, well, they just so happen to find that they deserve compensation.

Adobe trained one of their models entirely on content they had licensed for that purpose.

As for the attribution problem, my understanding is that the current generation of GPT basically _can't_ provide sources given limitations in the underlying technology (this is from some friends working on this at Microsoft). It's an ongoing challenge that they're not sure can be solved.

[Kickaha]

Originally Posted by Ryan Adobe trained one of their models entirely on content they had licensed for that purpose.

Well, of all companies to act ethically, Adobe was not on my list, HOWEVER... of all companies to be terrified of IP lawsuits by major IP copyright owners with deep pockets, they are definitely near the top.

As for the attribution problem, my understanding is that the current generation of GPT basically _can't_ provide sources given limitations in the underlying technology (this is from some friends working on this at Microsoft). It's an ongoing challenge that they're not sure can be solved.

You mean that they won't solve, because the dorks didn't consider it as a needed requirement. (Why would ethics be a requirement? How stupid, right?)

The current crop of LLM technologies are not capable of it because they chose not to make them capable of it. And now, they have so much $ and effort invested in the current systems that no way in hell are they going to ditch them and start over.

Much cheaper to convince the public (and buy legislators) that it's inevitable, that it's unavoidable, and golly gee, let's just live with it.

[Ryan]

Mozilla has spun off a non-profit to investigate ethical AI. If anyone is going to do it right it'll probably be them.

[Brad]

Timely news on the subject of unethical training:

Report: Potential NYT lawsuit could force OpenAI to wipe ChatGPT and start over

Weeks after The New York Times updated its terms of service (TOS) to prohibit AI companies from scraping its articles and images to train AI models, it appears that the Times may be preparing to sue OpenAI. The result, experts speculate, could be devastating to OpenAI, including the destruction of ChatGPT's dataset and fines up to $150,000 per infringing piece of content. ... Of course, ChatGPT isn't the only generative AI tool drawing legal challenges over copyright claims. In April, experts told Ars that image-generator Stable Diffusion could be a "legal earthquake" due to copyright concerns. But OpenAI seems to be a prime target for early lawsuits, and NPR reported that OpenAI risks a federal judge ordering ChatGPT's entire data set to be completely rebuilt—if the Times successfully proves the company copied its content illegally and the court restricts OpenAI training models to only include explicitly authorized data. OpenAI could face huge fines for each piece of infringing content, dealing OpenAI a massive financial blow just months after The Washington Post reported that ChatGPT has begun shedding users, "shaking faith in AI revolution." Beyond that, a legal victory could trigger an avalanche of similar claims from other rights holders. ... To defend its AI training models, OpenAI would likely have to claim "fair use" of all the web content the company sucked up to train tools like ChatGPT. In the potential New York Times case, that would mean proving that copying the Times' content to craft ChatGPT responses would not compete with the Times. ... Last month, the Associated Press became one of the first news organizations to strike a licensing deal with OpenAI, but the terms of the deal were not disclosed. Today, AP reported that it had joined other news organizations in developing standards for the use of AI in newsrooms, acknowledging that many "news organizations are concerned about their material being used by AI companies without permission or payment." In April, the News Media Alliance published AI principles, seeking to defend publishers' intellectual property by insisting that generative AI "developers and deployers must negotiate with publishers for the right to use" publishers' content for AI training, AI tools surfacing information, and AI tools synthesizing information.

[drewprops]

Ethical indeed.


...

[drewprops]

I'm going to be on a working group to analyze how AI will impact the film industry. We (here) already know that it's another disruptive technology, and that it will redefine how work is done across the spectrum as we move into the future.

I cannot simply have the fatalistic attitude of what will be, will be. But I need help envisioning legitimate means of combatting the elimination of jobs through technological advancement.

There is a lot of very valid fear out here about AI decimating the workforce of film technicians.

I personally believe the real threat to cinema isn't AI movies, but rather video games.

Visionary input would be greatly appreciated.

...

[psmith2.0]

Originally Posted by drewprops Then Paul asks his ethical AI to create a script that blends Star Wars with my film.

“Paul” me?

This Paul is outta the Star Wars business, I believe. And Indiana Jones, Marvel and all the rest.

I’m only interested in weepy Merchant-Ivory period dramas with ample 19th century cleavage, starring Gillian Anderson in her British accent, and giving the butler something good to look at!

“Computer, give me Dana Scully, barefoot, being chased around by Anthony Hopkins!”

Is that how it works?

[PB PM]

Originally Posted by drewprops I'm going to be on a working group to analyze how AI will impact the film industry. We (here) already know that it's another disruptive technology, and that it will redefine how work is done across the spectrum as we move into the future. I cannot simply have the fatalistic attitude of what will be, will be. But I need help envisioning legitimate means of combatting the elimination of jobs through technological advancement. There is a lot of very valid fear out here about AI decimating the workforce of film technicians. I personally believe the real threat to cinema isn't AI movies, but rather video games. Visionary input would be greatly appreciated. ...

The reality is that AI is going to hit the film and modelling industries hard. Why pay actors and models when the AI can generate the content without the cost, trouble with hiring, and the backlash that comes when actors and models do stupid stuff in their personal lives? It’s a no brainer for the companies involved. No need for film crews and photographers either. Increase profits for shareholders, that’s the CEOs job, if they don’t choose AI over humans they aren’t doing their jobs at the end of the day.

[psmith2.0]

I’ve thought this for a while. It’s the “business” part of “show business” taking lead. They’d be crazy not to. AI can’t OD on drugs, beat their wife/girlfriend, say something stupid that gets them tossed in cancelville for 6-10 months on a never-enough apology tour and atonement song-and-dance. Some will never come off it, depending on various factors. What studio/director needs the risk/hassle? They’d probably revel in the ability to take “stars” out of the formula. Some stars, anyway. The troublemakers and headache-givers. I figure in about two decades all major movies will be high-end cartoons…actors, locations, stunts, intimacy scenes. No AI actress is gonna wait 30 years to announce that so-and-so once grabbed her ass in 2013, and made an off-color joke/comment.

Remove actors from the equation and movie/TV show production probably gets real affordable, and easier. No hiring of multiple runners/flunkes whose sole job is to keep Ms. Streisand or Mr. Clooney supplied with some crazy, hard-to-find brand of sparkling water or cigars. Actors aren’t accidentally shooting/killing members of the crew, nobody’s getting raped or assaulted (or accused of it with no evidence).

[kscherer]

Originally Posted by PB PM The reality is that AI is going to hit the film and modelling industries hard. Why pay actors and models when the AI can generate the content without the cost, trouble with hiring, and the backlash that comes when actors and models do stupid stuff in their personal lives? It’s a no brainer for the companies involved. No need for film crews and photographers either. Increase profits for shareholders, that’s the CEOs job, if they don’t choose AI over humans they aren’t doing their jobs at the end of the day.

And the AI humans will all have perfect skin, and sleek, slender, muscular bodies.

The "fat is normal" crowd will go bananas when they're underrepresented by AI. So, AI will generate a lot of those Fat Bastard people to satisfy the offended.

Movies are gonna be so wo weird going forward.

But, hey, at least Jennifer Aniston's plastic surgery will start to look more "natural".

I can't wait for the first, feature-length film using actors/actresses/editors/etc. that are entirely fictional, and only cost 1.2 million to make but generate over a billion dollars. That will be fun.

[PB PM]

I’m sure there will still be smaller studios with humans in the process, with the selling point being, ‘look real people are in this movie/show!’. A list type actors only, the subpar wannabes will be on the street flipping burgers, if that’s even a thing anymore.

Then a big controversy breaks out when people find out even that studio only uses humans for the lead rolls.

[kscherer]

Originally Posted by PB PM I’m sure there will still be smaller studios with humans in the process

Yes, the studio will be called "YouTube".

And I've got dibs on that idea!

[PB PM]

Originally Posted by kscherer Yes, the studio will be called "YouTube". And I've got dibs on that idea!

Nah. YouTube is not the kind of stuff we are talking about here.

On a side note, YouTube is apparently putting some rules in place in terms of AI content on the site. Don't know the details.