Reply With QuoteI guess the people here who have been sharing AI-generated Thief artwork should now officially consider themselves cancelled.
How do you feel about this?
https://80.lv/articles/players-cance...t-controversy/
The tweet:
https://twitter.com/SystemShockGame/...45008777674779
"We will use AI again to create other pieces (including artwork). We may well use AI in other areas too. But this will never be at the expense of using skilled people or their creative talents."
I guess the people here who have been sharing AI-generated Thief artwork should now officially consider themselves cancelled.
I only play games that have a human hand-placing each pixel on the screen in real time as I'm playing.
The AI stuff is peanuts compared to how messy the whole thing has been. I backed the original KS and honestly I’m so over it I wasn’t even aware there was a new controversy. Again.
Such a waste.
How the taff does Night Dive keep screwing up? I remember they used to be at least decent once upon a time.
Is this Night Dive screwing up, or are people over-reacting?
There's nothing to see here if you ask me.
Prime Matter posted the AI tweet, not Night Dive.
And as far as I can tell from the article, one person claimed to have cancelled their preorder.
I think using AI art, even just for a marketing tweet, sucks. But this is a storm in a teacup.
I guess there is no going back when it comes to AI now, but the players or consumers do have a voice. If the community as a whole rejects content that is generated by AI (which is really just art taken from artists and used without their permission to train the AI) then they might back off.
The vast majority of human artists are derivative and rely just as much on existing artwork for ideas and inspiration.
I'm going to blow off work for a few minutes to expand on Jason's point.
About 100 years ago, people made animated shorts by hand drawing and photographing one frame at a time. Then somebody figured out that you could layer artwork on celluloid and didn't have to redraw it all every frame, you only had to redraw the cel that was changing. Then there were multi-level camera setups to automate the motion of objects over a background with realistic parallax for perception of scene depth. And using actors and rotoscoping to make human characters move realistically. The processes became efficient enough for Disney to make feature length animated films. As computers got more powerful and cheaper, animation started going digital in the 1980s and 1990s. Early digital animation required a great deal of effort and technology investment, but now a consumer can make animated shorts on an iPad to share with friends, using free art from any number of online repositories.
Snow White's production team had 32 animators, hundreds of artists, and a thousand assistants. The vast majority of them had no creative input and performed tedious, repetitive work. Consider what they were able to accomplish with 1930s technology, versus something like Cars (2006) which was made with about the same number of animators, fewer artists, and vastly fewer assistants. The latter is a vastly richer and more complex work thanks to computer automation tools, because they allowed the artists to spend more of their time exploring and expressing their ideas and less time on the mechanics of creating the work. When I was dicking around in the 80s with making simple sprite games on an Atari 400 and C64, I had to specify every pixel in code. It took me months to make my own Joust clone. Now a smart person like Henke can prototype a 3D third person game with physics in a weekend because of the tools.
AI art programs are just the next generation of content creation tools. They change the skill set needed to be a good artist, but they don't replace the artist, because they don't supply the vision and intent. They're going to enable development teams to make bigger, richer, more detailed, more realistic looking game worlds. They're also going to enable independent developers to make bigger and fancier looking games. If previous history is any indication, they're not going to put artists out of work. The advances in automating animation I referred to above only led to more artists and animators working in Hollywood.
Last edited by heywood; 10th May 2023 at 18:21.
Thank you for taking the time for a longer reply but I am not sure I can agree with you on all points. I think there is a big difference between an artist who takes a couple of images to form a mood board for inspiration, and letting an AI train on artwork from actual artists. I think a lot of people wouldn't have a problem with it, if in the end didn't involve making money off of it where the artists weren't compensated for. Many stock companies, for instance, want you to link to your reference image if you want to upload your hand drawn illustration for instance for that very reason.
I think in the end it will not amount in more jobs for artists but significantly less. Before you had a modeler and texture artist for the model, and now it is generated by AI, then you had an animator - now you have AI, then you needed environmental artists - now mostly generated by AI, and the dialog and story...guess what...generated by AI, with maybe one guy overseeing the whole development and adjusting things to his vision as you called it. Of course, we are not going backwards, but just as they learned later on in the automobile industry, that is is better to leave certain jobs to humans and not robots, the entertainment industry should learn that the human touch just makes for better entertainment. (Btw. back in the day I painstakingly changed the sprites for a C64 game called Falcon Patrol in the machine code as well to insert my own aircrafts ..haha..good times)
The question of AI-generated art is just the latest step in the larger question of how we as a society deal with automation in general.
In theory, automation is supposed to allow us to free ourselves of drudgery, to have to work less and have more free time.
But, in practice, it almost never works like that. Instead, it just ends up consolidating more money into fewer hands. The people who used to do the task no longer have jobs, and instead there's a smaller number of higher-paid people who maintain the automation.
Any saving of work is instead translated into growth and profit for corporations, and the people who no longer have jobs have worse quality of life as a result.
Unfortunately, most people don't get to do art for art's sake, they need to also do it to make money, because that's the nature of the society we live in.
I didn't expect a lot of people to agree with me, at least not now. It will get debated over the next few years and we'll see where we end up.
There's one point I don't get though. We teach art to people using artwork from actual artists. Why wouldn't we use the same material to train an art program? As with people, I think we should judge AI based on the quality and originality of its output, not its training pedigree. If all it can do is copy and composite, it's not going to be useful for producing publishable works.
Regarding the bigger picture, automation doesn't create long term unemployment. It creates short term structural unemployment. I don't think we've ever had a technology revolution that resulted in less jobs or lower standard of living overall.
It's not really the same, though. Human brains are fuzzy and (most of them) can't perfectly reproduce anything. They just take inspiration, they don't directly copy.
AI is more like if someone learns art by saving lots of other peoples' images and then makes new art by only copy-and-pasting pieces from their library.
The original artists likely won't be happy about their artwork being used that way without their permission.
Humans can cut and paste, and do it on occasion. And have you ever seen an AI art program spit out a perfect reproduction of another artwork? I haven't. I think that's a red herring. If somebody copied my work and published it as their own, I wouldn't be happy regardless of whether a person did it or a program did it or how exact the copy is. As artists start using AI programs to create paid works, they have to accept the risk that the AI might get them into copyright or trademark trouble if it spits out a wholesale copy of something and the artist doesn't catch it. There's probably going to be a few high profile accusations and cases, but I'm not worried about it because software that gets people in trouble isn't going to succeed in the market.
I'm excited about what AI can potentially bring to the table in gaming. So far, one excellent application we've seen is the generation of 3D models from satellite imagery for Flight Simulator. Another good application we've seen is upscaling. A few more obvious uses which are already in the pipe somewhere or imminent are text content generation such as dialogue and lore, voice acting, and art/texture generation. Given time they will be able to handle animation - the example videos being circulated now show there's still some work to do. I also expect AI will make its way into the tools for improving "automagic" things like lip syncing, merging intersecting geometries, aligning textures, and detecting problems. And once integrated into the level design tools, it won't take long before AI programs will be able to flesh out visual details in geometry and texture given some inputs about style or example to follow, because that's a straightforward extension of what they do now.
Thinking longer term, another part of the industry where AI could automate a lot of the work is play testing.
Correct. They're just sore that an algorithm is pumping out better stuff than they are.Originally Posted by heywood
It's one thing to be sore, yet another to lose your livelihood because companies and clients now choose to go the "get art and illustrations for free" route. There are many stories on reddit now about people having lost many of their customers because of this. Eventually, it will all sort itself out I guess, but to just embrace it without looking at the down side is a bit naïve I think.
Not dissimilar to the weavers who were smashing up mechanical looms two centuries ago. It didn't work out too great for them back then either. That's innovation for you.
You're a regular Edward Diego, SD.
Which, if the prior example of language translation is anything to go on, will result in artists being out of work long before AI is actually capable of producing comparable content.
See, I remember all these very same arguments when synthesisers were going to put traditional musicians out of work. And yet, still no shortage of people who want to listen to mediocre journeymen plucking strings and bashing things with sticks.
The state of Virginia was using machine translation for their COVID 19 website to help with vaccine scheduling. Among other things, it mistranslated the verb "to book" (as in "to reserve a date") and translated "The vaccine is not mandatory" to "The vaccine is not necessary". So at least when it comes to translating between English and Spanish, two relatively large and close languages, there's still a long way to go.
Maybe one day there will be a LLM-based AI that doesn't make such mistakes, but human translation is still necessary in any field that values accuracy and/or quality. There isn't really a market for machine-translated books, for example.
Also, while that may change with advances in neural network-based machine translation, right now it often takes less effort to translate something from scratch than it does to fix/edit a machine translated text, although companies more and more seek to cut corners by offering post-edited machine translations instead of human translations, due to editors generally having a lower pay grade than translators.
If your argument is 'in fact, there's place in the world for both AI generation and traditional artists with some necessary turmoil and displacement before equilibrium is reached', that's a far more optimistic point of view than I'd otherwise ascribe to you.
So I'm going to assume what you actually mean is eventual artistic subsumption is better overall because it'll consistently match some arbitrary qualitative parameters you and people like you set, which is definitely a take.
Well, it's interesting, right. I don't think it's exactly a red herring because there's more nuance to the discussion than 'that's what humans do too'. AIs are in some ways instantaneously more capable than someone setting out to learn from the masters, and there are in fact not that many humans who have that predilection to begin with out of the entire population; this reinforces a certain inherent rarity to the art that we produce. Put simply, this means that inspiration and derivation are an acceptable process because it results in a certain amount of artistic value, where deviations from the originals filtered through the artist's mind and life (their journey, to reference a conversation in Comm Chat) to a certain skill level contribute to the work having more value.
The fact that you can make as many instances of these ANNs as you can afford hardware to host them on means that we're introducing two problems: one, your population of 'artists' increases drastically because you can make them at will, so to speak, which decreases the rarity factor. Two, since AIs have no actual life experience or even have 'understanding' as such, they will deviate from copyrighted art only inasmuch as they can reconfigure it in ways that amount to outright copying or pastiche, which are traditionally less valuable artistically. We don't have issues with humans doing this, because there simply aren't that many humans out there who will. Putting both together: when you can configure an army of the things, limited only by the amount of hardware you can host them on, you can see why an infinite pool of art permutations from copyrighted sources is problematic to the creators - this can have the effect of devaluing original works from less famous artists, simply because we will have an incredible amount of stuff to sift through if someone chooses to flood a market.
We're already seeing this a bit with Spotify and its wave of 'artists' with generated music, and while we're not at the stage where it's difficult to confuse them with 'name' artists, I don't see a future where sifting through all this chaff isn't going to be extremely difficult - especially given that, with enough time and training, we'll reach the stage where competency and variability is good enough that we may simply not care if it's a human or a robot that made it.
Some very interesting takes on this. It certainly makes one think about values and what it means for us going forward. As I said earlier in this thread, there is no going back. One just has to adapt to the new reality. For certain jobs it will be devastating but for others it will open new opportunities. It could even be that we get some sort of content labeling in the future similar to the Non-GMO food labels.
Not sure where the unbridled hostility is coming from, but it's essentially true that artists with talent have nothing to worry about from artificial intelligence. At least until robots start picking up paintbrushes.
Posting Permissions