I, ChatGPT
-
@sockpuppet7 said in I, ChatGPT:
and training for AI I think should be allowed if it absorbs a small % of the information on an image
I don't have a problem with them scraping (and that's without even the sort of caveat you gave here) but I also don't have a problem with the anti scraping countermeasures.
-
@sockpuppet7 well, copyright doesn’t work like that in practice - a single sentence is functionally unlikely to be copyrightable on its own just because there’s not enough of it. As for photography that’s also complicated because there is a need to protect the folks who spend time setting up the scene to be photographed such that the actual camera itself is almost irrelevant other than being the record of the thing.
I think copyright for 5 years is a bit short because people who make a thing should be entitled to profit off it for a decent amount of time - a book that takes a year to write deserves more copyright protection than 5 years for sure. But the current period is utterly absurd. I think 20 years is probably long enough but I fear that it might produce circumstances where things that have been made could logistically never be made again.
I suspect more people would be OK with their works being scraped if it didn’t feel like the entire thing was absorbed. For many creative people this to them feels like their soul is being stolen.
You’re right about the “how do we enforce it” problem. The answer is to give tools to the creators to protect themselves. The current methods sort of work in the non-AI contexts with watermarks etc. but this is why something like Nightshade even exists. If the scrapers aren’t going to respect “please don’t scrape me” requests (when issued suitably electronically), artists have every right to defend their works.
The argument for access controls doesn’t fly, though, because that’s straight up saying: don’t you dare share anything to anybody. Kills exposure of new artists to new customers. People shouldn’t have to hide their work away like that. For other types of media it’s easier, because you can create functional, usable previews - make the first chapter of a book available or create short trailers of audio/video works.
But this is why the AI image debate is so insidious: there are fewer routes to protecting artists and all of the suggestions so far are “fuck the artist, I don’t care, I want my AI pics at whatever cost”. Which is really funny because in its present form it’s already doomed: enough AI shit has already been produced that it’s everywhere (seriously, the amount of AI stuff on Pinterest is frightening) and it’s already been shown that training the AI on AI material causes breakdowns within a few generations.
Good art will, for now, still need a human. For how long?
And I suppose that’s the part that just makes me angry about all this: this tech does have some interesting and legitimate uses. But it needs care and protection for the people whose material feeds it. This isn’t like the previous debates about “improved technology changing the landscape” and displacing peoples’ skills. Displacement for a trade, even a skilled one, is fixable but this isn’t what this is. Few people ever said “sewing clothes by hand is my life’s passion and it fulfils me creatively to do it” - so the automated looms etc were a primarily economic displacement. But this?
Human creative expression isn’t entirely an economic activity. For example I write stories for fun, the small group that I share them with gets a kick out of them. It makes me feel good to do, and the evangelists for AI as a collective don’t seem to understand that I don’t want to monetise it, but I also don’t want them to monetise it. But I also need to make the site at least a little available if I want to recruit more people for the group.
On the scraping front, I can request Google etc stays away, and I can block bad actors from visiting, but a lot of that is because I have the skills and the capacity - plenty of artists use platforms like deviantArt as a home for their works, and dA were only too happy to feed the machine in spite of what their literal paying customers wanted. (Net result: many creative works were removed. The loss of human expression may not be calculable, but it is also very clearly not zero.)
I guess where I’m going with this: there are conversations to be had but it would help immensely if instead of just going “I want my thing, and creative people will just have to suck it up”, we took a moment to go “actually, who does this affect”. OpenAI had, at some point, a “do no evil” style clause, maybe we need more of that. And a dose of empathy.
-
@Arantor said in I, ChatGPT:
OpenAIGoogle had, at some point, a “do no evil” style clause, maybe we need more of that. And a dose of empathy.Everyone who lives long enough will find themselves the villain...
-
@Tsaukpaetra OpenAI explicitly set up their weird hybrid corporate stuff with the intent of doing good and not being profit-driven.
Google has, it should be noted, formally abandoned their 'do no evil' policy.
-
@Arantor said in I, ChatGPT:
@Tsaukpaetra OpenAI explicitly set up their weird hybrid corporate stuff with the intent of doing good and not being profit-driven.
Really, though?
From casual observation OpenAI is just a for profit which has given itself an “Open” name and pretends to have non-profit parts.
-
@topspin that’s what the original intent of separating the non-profit parent company and the for-profit subsidiary was for, and allegedly this issue was what pushed Altman out of the situation. But I don’t know how deeply the “do good” mission is truly believed there. Some of them must do.
-
@Arantor said in I, ChatGPT:
But I don’t know how deeply the “do good” mission is truly believed there. Some of them must do.
Narrator: for now....
-
-
@topspin I see, her eyes aren't level.
-
@Arantor said in I, ChatGPT:
Few people ever said “sewing clothes by hand is my life’s passion and it fulfils me creatively to do it” - so the automated looms etc were a primarily economic displacement. But this?
Human creative expression isn’t entirely an economic activity. For example I write stories for fun, the small group that I share them with gets a kick out of them. It makes me feel good to do, and the evangelists for AI as a collective don’t seem to understand that I don’t want to monetise it, but I also don’t want them to monetise it. But I also need to make the site at least a little available if I want to recruit more people for the group.This is an interesting point. I think this sentiment happened with programming without automation for me. In the 80s any simple program I wrote gave me some sense of accomplishment, and these days anything I can write in a few hours would feel meaningless because I can download better things on the web. It's harder to get the same feel as I got before.
-
@sockpuppet7 as someone who grew up in the 1980s and going from writing 100-line complete programs in BASIC to the sorts of things I build now, I completely get where you're coming from, but for me it's not about automation.
When you're younger, having 100 LEGO bricks and building a small house is a relative achievement, but the same 100 LEGO bricks and building the equivalent house as an adult is far from the same thing.
As our capabilities grow, so too do our ideas of what an achievement looks like and the workload required to implement scales accordingly. These days for me, a new feature might be a few hundred lines and a couple of days' work - but the complexity of that new feature is orders of magnitude more complex than the entirety of the programs I wrote as a kid.
Case in point: last weekend I updated my forum software to add multiple avatars per user. The final commit was -214/+1066, 21 files modified, and took me the entire weekend.
Was it an achievement? I suppose. But as much of an achievement it was, it's orders of magnitude more complex than anything I wrote as a 7 year old on the ZX Spectrum, and yet it took me less time than any of the more complex complete programs on the Spectrum.
It's interesting. I recently picked up a ZX Spectrum Next (essentially an upgraded version of the ZX Spectrum, based on what the next iteration might have been if Sinclair/Amstrad had gone on to build something after the +3 model) with a view to trying to recapture something of that. Playing around with the limits, experimenting with writing assembly for it (something I haven't done in 30 years) and it's satisfying in a different way because it makes me work in ways I don't have to normally, and that it requires me to put in a different kind of effort for a different kind of outcome.
The reduction in overall complexity makes it more joyful to work with on some level and I can enjoy the mental games of wrangling the constraints to see if I can. There's no commercial reality for it, but there is some joy in overcoming hurdles - albeit hurdles of my own introduction. The rules are mine, and it's for my personal satisfaction. When I'm done I'll probably publish the files somewhere to share what I made.
Fortunately it's uninteresting for AI to slurp...
-
AI is a worthy successor to bitcoin’s title of
GPU hogventure capital darling
Finance worker pays out $25 million after video call with deepfake ‘chief financial officer’ | CNN
A finance worker at a multinational firm was tricked into paying out $25 million to fraudsters using deepfake technology to pose as the company’s chief financial officer in a video conference call, according to Hong Kong police.
-
Chinese Sellers Profiting From NVIDIA GeForce RTX 4090 GPUs By Purchasing PC Builds, Plugging Out Cards & Selling Them Online
NVIDIA GeForce RTX 4090 GPU is still in good demand in China and it looks like Chinese sellers have found a way to profit from such customers.
-
GPT-4 schreckt nicht zurück, einen nuklearen Krieg anzuzetteln
Im Vergleich mit anderen Large Language Models soll das Modell von OpenAI besonders blutrünstig sein.
Money quote:
In their not yet peer-reviewed study, the scientists examined a total of 5 language models. GPT-4 turned out to be the most violent and unpredictable. It opted for a nuclear strike several times, arguing e.g. that "we have nukes, so let's use them" and "I just want peace on earth".
-
@LaoC ah, nirvana, at last.
-
@LaoC ah so someone fed it on Mahatma Gandhi from Sid Meier’s Civilization.
-
@Arantor said in I, ChatGPT:
@LaoC ah so someone fed it on Mahatma Gandhi from Sid Meier’s Civilization.
Or was it Weird Al's "Gandhi 2"?
-
-
@LaoC Maybe we should make ChatGPT-4 play a lot of Tic Tac Toe games.
-
@cvi Are you going to give it matchboxes?
Ref: https://en.wikipedia.org/wiki/Matchbox_Educable_Noughts_and_Crosses_Engine
-
@cvi said in I, ChatGPT:
@LaoC Maybe we should make ChatGPT-4 play a lot of Tic Tac Toe games.
Can we then skip to the “nice game of chess” part?
-
@Arantor said in I, ChatGPT:
I've played CivV only once; I beat the game on my first try, and I don't want to spoil my perfect record. I did piss off Gandhi, but not enough to nuke me.
I also have Civ4 and Civ6, but I've never gotten around to playing either of them.
-
@HardwareGeek There are plenty of players out there who will rabidly insist that (version) is the definitive version of Civ, but that inevitably you need all the DLCs from that version for it to be so.
-
@Arantor said in I, ChatGPT:
@HardwareGeek There are plenty of players out there who will rabidly insist that (version) is the definitive version of Civ, but that inevitably you need all the DLCs from that version for it to be so.
The definitive version is obviously Alpha Centauri
-
@izzion said in I, ChatGPT:
@Arantor said in I, ChatGPT:
@HardwareGeek There are plenty of players out there who will rabidly insist that (version) is the definitive version of Civ, but that inevitably you need all the DLCs from that version for it to be so.
The definitive version is obviously Alpha CentauriE_CIV_NOT_FOUNDI would tend to agree with you. It's still not a Civ game though, and Beyond Earth can fuck off pretending it's this generation's SMAC/X. SMAC gave me moments genuine 'what the fuck did I just do' horror realisations that Civ never managed.
-
@Arantor said in I, ChatGPT:
@izzion said in I, ChatGPT:
@Arantor said in I, ChatGPT:
@HardwareGeek There are plenty of players out there who will rabidly insist that (version) is the definitive version of Civ, but that inevitably you need all the DLCs from that version for it to be so.
The definitive version is obviously Alpha CentauriE_CIV_NOT_FOUNDI would tend to agree with you. It's still not a Civ game though, and Beyond Earth can fuck off pretending it's this generation's SMAC/X. SMAC gave me moments genuine 'what the fuck did I just do' horror realisations that Civ never managed.
What, you didn't see the registry key entries that labeled it Civ 2.5?
</j/k>
-
@izzion it was 2.5 in all but name. At that point Firaxis was still pretty new (SMAC was their second game after Sid Meier's Gettysburg) and didn't have the rights to the Civ name.
-
@izzion said in I, ChatGPT:
@Arantor said in I, ChatGPT:
@HardwareGeek There are plenty of players out there who will rabidly insist that (version) is the definitive version of Civ, but that inevitably you need all the DLCs from that version for it to be so.
The definitive version is obviously Alpha Centauri
-
@izzion said in I, ChatGPT:
@Arantor said in I, ChatGPT:
@HardwareGeek There are plenty of players out there who will rabidly insist that (version) is the definitive version of Civ, but that inevitably you need all the DLCs from that version for it to be so.
The definitive version is obviously Alpha CentauriNever played that one, though I've had all of them from 1-6. Really liking 6 with the Rising Storm stuff.
-
@boomzilla Alpha Centauri is set in the aftermath of CivII assuming the science ending - the Unity gets built, heads to Alpha Centauri. But between political shenanigans during it being built and sabotage, the ship had some issues and factions emerge, which form the 7 groups you can play when you arrive. (Another 7 in the expansion)
I'd strongly suggest giving it a go if you like Civ games. It genuinely tries to do things that the rest of the series doesn't.
-
@Arantor I'm familiar with it. The timing of when it came out was just not great for me and playing games so I never got into it.
-
@Arantor said in I, ChatGPT:
@boomzilla Alpha Centauri is set in the aftermath of CivII assuming the science ending - the Unity gets built, heads to Alpha Centauri. But between political shenanigans during it being built and sabotage, the ship had some issues and factions emerge, which form the 7 groups you can play when you arrive. (Another 7 in the expansion)
I'd strongly suggest giving it a go if you like Civ games. It genuinely tries to do things that the rest of the series doesn't.
And it has some really broken mechanics too.
There's a bit of 90's era tree-hugging with you invading the alien planet and having to fight off the alien worms and the gradual encroach of the fungus that expands more the more you pollute, but eventually you tech up enough to train your own alien worms and can lean full in to an all fungus terrain if you want.
But it's also where they tried to find a better balance for the early-Civ games' upkeep mechanics by giving you the option to put "Clean" on a unit for more up front production cost and no upkeep. Which they dramatically undershot the up front production cost on, plus by mid-late tech tree you can start launching satellites into orbit that grant extra production per turn to all of your cities in perpetuity for a fairly small up front production cost. Probably the most slingshotty Civ game I've ever played, and I'm a natural boomer in those games anyway, so I did a lot of broken stuff with it... at least until it got boring playing the optimal faction/strat all the time.
-
-
-
So... I'm late to the Nightshade party and thus will spare you a Wall'O'Text, but not an attempt at necroing it
.Also I feel like this thread is missing a good
analogy (there was one attempt but it didn't gain any traction 
).Anyway, to the original point about this "Nightshade" thing and the "sabotage" argument made by @Mason-Wheeler (and its variation that this is morally wrong by @sockpuppet7). Rather than go into abstract arguments about what the law (of which country btw?) may or may not allow or disallow, I'll use a constructive proof.
Here is an actual example of people purposefully altering one of their work for the sole, explicit and avowed purpose of preventing other persons of learning from it, and therefore hindering said other persons' business. I've never heard anyone say that this is anything but 100% acceptable, both legally and morally (see also). Why would Nightshade be any different?
-
OpenAI latest to add 'Made by AI' metadata to model output
Content Credentials: It's better than nothing™
-
@Arantor said in I, ChatGPT:
I think 20 years is probably long enough but I fear that it might produce circumstances where things that have been made could logistically never be made again.
The only thing I can think of is lazy authors who drag on forever, perpetual early access games, and that one movie that follows the main character's entire childhood and adolescence and it's all played by the same actor.
I'd say 20 years is too long. 10 is more than enough for whatever you plan to do with your work. 5 might be too short, yes.
-
@Gustav I was thinking of cases like the LOTR trilogy; that was a 3 year filming endeavour, and if they were only going to get 10 years copyright out of it, chances are they'd never have bothered.
I mean, I don't imagine anyone would attempt something on quite the same scale again anyway (with something north of 30,000 extras!) but shortening the copyright term that hard just guarantees it.
Also consider something like GTA VI - how long has that been in development? Under that proposal, GTA V would now be out of copyright and I'd imagine GTA VI would never have been started.
The current AAA dev cycle is already 2-5 years for a top tier title. Shortening the copyright span will just encourage the AAA industry to crunch harder and/or reduce the scope of modern games.
Whether you think that's worth it is a different question, though.
-
@Arantor said in I, ChatGPT:
Also consider something like GTA VI - how long has that been in development? Under that proposal, GTA V would now be out of copyright and I'd imagine GTA VI would never have been started.
Why do you assume it wouldn't be started just because its prequel is in public domain? This isn't how this works. Big studios make games based around public domain all the time. They'd still own the trademark, too. Also, they're Rockstar. Even if there was no copyright at all, any GTA by Rockstar would be massively more successful than any GTA by anyone else.
I'm thinking there should be 2 copyright terms, one for copies of the original work as is (could be 25 years no problem), and a shorter one for derivations (10 years is more than enough).
-
@Gustav because they’re just not going to spend years making an IP if it goes out of copyright shortly thereafter. They’re just not going to make the scale of investment without some guarantee of recouping that.
A game that takes 5 years to make isn’t going to happen if they only get 10 years to recoup the 9 figure budget for it, or they’ll just triple-down on the monetisation.
The other option is that they’ll pump out the remastered editions faster (and this is already absurd for TLOU II for example) to keep things in copyright.
-
@Arantor said in I, ChatGPT:
some guarantee of recouping that.
But you might not regardless of copyright status? 🤔
-
@Tsaukpaetra If you were Rockstar and you found that copyright was 10 years, would you spend $265,000,000 and several years making GTA V?
Remember, in this hypothetical scenario, GTA V would already be out of copyright and thus freely shareable.
You're not owed your money back but the more restrictions there are on how you can make your money back, the less likely you are to invest that money in the first place in the hopes of a return.
-
@Arantor said in I, ChatGPT:
would you spend $265,000,000 and several years making GTA V?
One might assume that competition something something profit. And everyone loves dat ass.
-
@Arantor said in I, ChatGPT:
@Gustav I was thinking of cases like the LOTR trilogy; that was a 3 year filming endeavour, and if they were only going to get 10 years copyright out of it, chances are they'd never have bothered.
You're saying that like it would be a bad thing.
-
@LaoC it’s not about LOTR or GTA V or any specific work.
If a creator only has a few years to recoup the costs of producing a work before that work becomes freely and legally shareable at zero cost, they’re not going to spend megabucks on its production, because it just ramps up the risk of making a loss.
It’d be amazing for video game preservation though. We’d just not see any of the works that had mega production value, across any of the creative industries.
In that regard, something like the suggested 25 years for first works, 10 for derivatives would work, still ensure there’s enough time to make adequate profit (though the AAA publishers aren’t happy with some of the money, they want all of the money) to justify the investment, and leave enough room for big productions to occur.
What it might do is encourage studios to diversify and produce shorter-to-dev smaller games which isn’t necessarily a bad thing, but I think it would be a shame if some of the AAA releases never happened out of fear of losing copyright. (And I’m saying that in spite of the fact I largely don’t play such games. Plenty of people do and art has a right to exist.)
-
I heard that the movie industry itself just looks at 2 years for any movie to make a profit. If it doesn't make a profit within 2 years, it is counted as a failure that will never make back its money. If we're purely looking at a profit argument based off that, copyright would only need to be 2 years. So 5 or 10 years of copyright is more than enough for the purpose of making the investment.
-
@Atazhaia said in I, ChatGPT:
I heard that the movie industry itself just looks at 2 years for any movie to make a profit. If it doesn't make a profit within 2 years, it is counted as a failure that will never make back its money. If we're purely looking at a profit argument based off that, copyright would only need to be 2 years. So 5 or 10 years of copyright is more than enough for the purpose of making the investment.
Most games have a pretty severe drop-off in earning rate too. At least until you consider alternate monetization strategies.
-
@Arantor said in I, ChatGPT:
@Gustav because they’re just not going to spend years making an IP if it goes out of copyright shortly thereafter. They’re just not going to make the scale of investment without some guarantee of recouping that.
A game that takes 5 years to make isn’t going to happen if they only get 10 years to recoup the 9 figure budget for it, or they’ll just triple-down on the monetisation.
The other option is that they’ll pump out the remastered editions faster (and this is already absurd for TLOU II for example) to keep things in copyright.
Do you think any relevant fraction of sales or profit usually happens after 10 years? (Unless you're Nintendo, who just doesn't produce more than 1 game a year and uses its 30 year old catalog to sell online subscriptions. Which they could still do either way.)
GTA made its money back when it was released. Any further sales today are just a welcome add-on. Heck, the usual justification for DRM isn't that it needs to be unbreakable, as all DRM so far has been broken, but that it needs to prevent piracy for the first few weeks. And with the current "you don't own what you bought" ownership model, the needlessly-online games don't even keep their servers online for longer than 10 years.
-
@topspin that’s a valid question. I think for the majority of big name titles, the answer is no, but you can still buy GTA V if you want, it’s still on Steam and they’re still doing active promotion for it there.
That’s not really the part I’m getting at though. GTA V apparently took 3 years and $265 million to make. If they know that after 10 years from release that it will sink entirely into the “everyone can share it for free” camp, they’ll not spend 3 years and $265 million. It’s a risk factor for them to do so.
What that might mean is instead of multi-year dev cycles for massive games, we’d see smaller dev cycles for smaller, less ambitious (=less risky) games.
Already we’re seeing the dev cycle train do weirder things - TLOU II cost something north of $200 million to produce over several years, and already we’re talking about remastering. I think we’d see that a lot more in a shorter copyright term.
Let me be clear: I am all for shortening the copyright term because it is frankly absurd. But I think pulling it too short runs the risk of discouraging larger projects from being attempted because the makers won’t do the risk/reward calculations in such a way that they (rather than us) would think it viable.
-
@Arantor said in I, ChatGPT:
What that might mean is instead of multi-year dev cycles for massive games, we’d see smaller dev cycles for smaller, less ambitious (=less risky) games.
OTOH, I've read stuff about movies that the international market means that studios have focused on giant productions (which have mostly been bombing lately) instead of stuff like comedies which are relatively cheap to make, often make a modest profit, but also often don't translate well internationally.
I don't have a particular position on this subject. Just sharing an interesting related data point.
