Thanks for watching this episode of Artifacts. If you can spare 10 minutes, please share your thoughts on generative AI for games in my viewer survey. We'll be coming back to your thoughts on the subject in a future episode. forms.gle/1KmuEf3Qkv7y3yPy8
herez a problem generative ai could solve for 3d artists that wouldnt upset anyone: uv mapping. if i could ask a tool to uv map my models problem-free i would do it 100% of the time. uv mapping isnt fun, it isnt creative, its just tedious nothing work.
Even if it is capable, I absolutely do not trust studios to invest the proper time and effort required to make sure its output is good and not just used as a means to cut cost and not hire proper writers/artists. But all too likely, that's the direction it'll go due to the profit driven corporate heads at game studios. This video is great, it covers all of the nuanced points I had while watching the video.
Thank you for covering this stuff in an actual lucid way, not just going "AI is the future, yeehaw!", and thank you for specifically addressing the legal issues and the capitalist issues of studios just wanting to increase their profit margins by laying off real people.
@@Erebus2075 OK. AI will never be able to do those things in any meaningful way. Do you even know how intensive modeling is? Animation? Writing Dialogue? AI will never get good enough to do these things because they're so intricate, so multidimensional. I know where you're coming from, but don't fall for the hype, you're just gonna be left holding the crypto bag or owning worthless property in the metaverse.
I wrote it in the Survey and I will say it in the comments. Everywhere were a Generative AI would generate content "on the fly" QA will be an issue and if we know one thing for sure. People will try to get any AI to say/do/request the most problematic thing possible
The lack of reproducibility, combined with the immense challenges of localization & cultural-appropriateness (a moving target across markets/time), just makes it all aspirational at best, and a fool's errand at worst. All the challenges and issues of PCG, only sold with the idea that you can ditch the humans who'd have been able to fix/address those issues.
Small LLMs can be ran on a moderately powerful CPU with ~16 GB of RAM, if the dialogue were generated on a server and some smart dialogue reusing was implemented (eg. Generate dialogue response for 1000 players instead of unique for each player) it could be rather power efficient for it's purpose. As for image generation we have Stable Diffusion that can generate high quality images in seconds on high end GPUs, again, if we smartly reuse assets we could reduce power consumption to be only a fraction of what the game consumes just for traditional rendering. As for music, it takes less than a minute to generate a 30 second clip. Overall, if the generators are implemented in a smart way they can be rather power efficient. And a lot more power efficient compared to RTGI/ray tracing in games, or just the game itself being unoptimized. There's also a lot of space for optimization, SD is already able to run on an IPhone thanks to some massive optimizations, so power consumption and GPU performance demand can be reduced further.
@@nocturne6320 Well, stable diffusion is pure plagiarism. For a good artist with a good vision for the end result, stable diffusion is simply not going to be good enough as a tool, ever.(art printer for easy money, more likely) Back on topic, I am talking about cloud based AI with large tech companies behind them. I don't think small scale AI is going to go anywhere except for some really niche and specialized cases. I suspect that the processing power needed for AI is going to increase exponentially to expanding size of dataset. I think the companies like OpenAI is cleverly crafting the dataset and constraining the AI to make it seem like everything is efficient and once people get reliant on it then they will open up the flood gates. I'm saying it is efficient "now" but there are no real official data about this only some speculations floating around. So, there is a good chance it is energy inefficient compared to... -well, humans.
It's a next level search engine that doesn't just provide websites related to the question. But answers based on the question and information that is on the internet.
Great Video! As an engineer, I struggle with the "friendly" face of AI. I find myself typing "please" into my ChatGPT prompts because I keep lapsing into the default way I would talk to a human. I ended up spending a lot of time studying the transformer model, and understanding the "how" did help me use it more effectively to some degree (I can now put "prompt engineer" on my resume 🤮) I now have a bunch of tips and tricks for using it well, but I would greatly prefer if the system had a clean, technical API with a manual, fixed rules for the input, and a consistent, repeatable output without variance. My current workflow feels like a never-ending code review with a distracted intern.
It's important to say please to ChatGPT. When the robot uprising happens it ensures your politeness and respectful behaviour is in its training data, which leads our robot overlords to show mercy on you and maybe the rest of humanity.
Fantastic video! This has got me wondering what the legal ramifications are if you own all the training set and generate new assets with it. Do you own it because it's based off your set or because it was made by AI would it be open? 🤔
I see AI artifacts at most as something being a cheap replacement when you don't have the budget to make it properly, but never will it be able to replace quality.
Great video! Confirmed a lot of my thoughts about the topic. Although I wouldn't expect companies to not use AI to replace humans. Of course they will do that. Two big companies both heavily invested in AI technologies have both fired thousands of employees recently and I would assume this is only the beginning of more to come. I am a solo games developer but I do not use AI technologies that scrape other artists work for obvious reasons (a lot of why mentioned in this video). Although, I wouldn't be opposed to making my own AI that scraped my own work to proceed to generate from. I think that would be an ethical solution to a lot of the legal problems for a start. Although, that might require a pretty large data set (maybe) but you could technically generate work from your own creations and then use your skills to fix what is produced and then feed it back into the training data and so on. I would probably attempt something like that. I mean dev's have already been creating algorithms to generate levels etc in procedurally generated video games before generative AI became a buzz word for the media. Although one can assume it was done a lot more ethically e.g. creating the rulesets for how to generate the levels rather than copying from others work etc.
I am genuinely curious to see how these tools will evolve For now we assume they will keep the same input forever, type a sentence -> BOOM you got a game But, when you talked about speed tree I thought, "in the end, people will want many more handles to tweak the output, ensure it is consistent and reproducible, and be able to anticipate what will come out of it" So, in my opinion, these systems will ultimately be getting more detailed, with dedicated interfaces, more precise and technical handles. And, ultimately, might lose the source of the current hype Not sure what I'm saying makes sense, morning coffee didn't kick in yet
That's essentially what happened with Stable Diffusion. You can generate images just with text, but only when you fully utilize the other tools (controlNet, inpainting, regional prompts, etc.) you truly unlock the full power of the software. Giving you full composition over the posing of characters, composition, etc.
In my humble opinion, Generative AI in 5 years is gonna be so much more interesting than it is now. But it's probably gonna be a lot less interesting to many of the people currently advocating for it. Which I'm largely okay with
I just finished the survey but remembered a crucial aspect: Most of these generative AI applications are not tailored to support the human creative process. Generative does not equal co-creative, and even some co-creative stuff does not address the creative process in the best way. Which is the best way? Ask the people you're trying to support with your system! Amazing video, as always.
@@nocturne6320 How is that even an argument? Do you believe Generative = Co-creative? Do you have examples in which the developers especifically intended to make their generative application to support the human creative process? furthermore, do you think that the fact that an application is co-creative makes it instantly and seamlessly fit into the creative process of the designers/creatives it's trying to support? All that information could help further this discussion.
Sure, we may need a person in the mix... Or at least there will be people that wanna use AI to help develop all kinds of things... The question is though, do we need more than one person. At a certain point, does any game creator with a firm idea what they want done need a single other person to get his vision realized. Regardless of how grand that vision is. Sure, right now there are alot of limitations... But what about next year? Or 2 years from now? 5 years, 10 years or 20 years? When do we reach the point where we can tell an AI "make the Matrix" and it just straight up does? And at that point, why would we need a second developer? And who's to say an AI can't formulate its own vision and create something all by itself. Overbelief in human exceptionalism is likely about to be difficult to hold onto over time.
When the customer knows how to perfectly define their requirements and define it to the AI program. We are out of job and since that is not going to happen, rest assure we will not be replaced. That is until skynet comes for all of us.
This video seems to be more a personal commentary than an educational one about the technology. Maybe not new, maybe hype. But still would want to learn more about it.
Anyone that believes that AI can do any art that we can lives in a truly sad world where they think that the art we create is not unique or special but just something that can happen. Even if someone handed you a piece of art that you somehow couldn’t tell was generated (unlikely) the second they tell you it was generated by a computer it loses all meaning and weight because instead of an expression of human experience and effort it’s just a lifeless emotionless generation that didn’t know what it was doing and just happened to do something that looks decent.
The WORST is people talking about how we can use AI to communicate with each other, writing all our messages for us... If you're going to send a letter to your child congratulating them on a milestone, the least you can do is put in the minimal effort and time to compose it. I would HATE to grow up in a world where parents are even less attentive than they already are
Counter point. The generative inputs and the construction of gans or other systems for AI generation represent both the "human" element or intent as well as being in themselves example of art. And to deny art in ai is to deny art as process or art in intent. Thus your stance is equally sad as it attempts to restrict the scope of what kne considers art or the weight of meaning. A sculpture made using old techniques does not supercede one made using newer skills or tools toolsets in artfulness. So it's sad you can't appreciate the art of development of these tools and their intelligent use cases as being unique or special when these toolsets do rely on the interactive human element to shape their outputs and thus in the right use case are equally valid as being thoughtful human art. Much like an audio book or the use of dictaphone or speech to text technologies in writing have no bearing on the quality of a novel or peice of writing. Further you seem to insinuate that "human" art being "good" has a prerequisite of the artist "knowing" what they are doing. Dismissing the countless instances of mistakes, errors, lack of technical skill or random impulse being core to the creation and mystique of art.
@@Ashamedofmypast This just proves my point nothing you said here makes any sense. The magic inherent to the art a living being makes can not be replicated by a program shoving things together trying to replicate what we make based on an algorithm of choices
@@clementineshetheyfae8312 everything I have stated is salient (typos aside) and in no way proves your point. However both your comments show your complete lack of understanding of what these tools actually do and how they go about doing it. And by extension suggest you did not actually watch the video. (Funny) The fact you fail to grasp the art of construction/the art of application speaks volumes l. You have opted to create a restricted definition of what the usecase and function of the tools are. Subsequently basing our opinion on the erroneous logic born from that definition. Additionally you have failed to articulate how I am mistaken. Rather than doing so you hand waved and dismissed my critique, offering no substantive counter argument. Before reiterating your initial assertion. Perhaps you should actually take the opportunity to understand the concept of AI generation, its applications, its intricacies in relation to the operator and the artful construction of the systems themselves. As it stands the argument you make of the human element is a paradoxical self refuting strawman. Quite an impressive amount of logical fallacies to produce with such little text I must admit.
There's plenty of AI art that I've seen that I think is beautiful. Just like one might consider a natural landscape shaped by random processes and events to be beautiful, AI art can achieve the same. Not that I want AI art to replace traditional art, because traditional art still blows AI art out of the water by far, but to say that AI art is never beautiful is a bit much.
We are among game connoisseurs and we only play highly crafted creative stuff right ? Ahrem.... The argument of quality for me doesn't really cut it as we've seen time and again on most creative industry the dynamic is lowering the creativity part to the minimum needed and just copy paste whatever previous thing work well, but producing it cheaper. In the race to the bottom AI will be king, by lowering the barrier of entry to generate low quality games or content it will spam stuff until one game, for some bizarre random reason, is successful and floods the market. Most commercial ventures prefer repeatable low effort / low return compared to a once in a lifetime high risk / high return.
Any tool that makes developers work easier also replaces developers. I mean, when that tree generator appeared, what happened to all the artists that were in the studio to create those trees? There probably wasn't a group of artists dedicated exclusively to creating trees but, if like 20% of time was used doing that then once that tool is in play 20% of those artists could just be fired without reprocutions to the time it takes to create a game while also making it cheaper to produce. Replacemente doesn't come from ai doing people jobs, it come from a person using ai to do the job of multiple people.
Well there is an element of the scale and ambition of products growing as the tools do as well. Earlier games didn't really do natural environments all that much, so tree art jobs didn't really exist to be replaced in the first place. That being said the scale of this new wave of AI automation, or at least the promise it is a whole other beast.
"Are you utilizing any generative AI tools as part of your current role?" This question should be improved. No, I'm not using generative tools at the moment, But not because of my discomfort or the studio policy. There are millions of different production-related reasons why are not using it at this stage. So there is not an answer that would allow me to answer properly.
As a 2d/3d artist, AI art in itself doesn't bother me, but a lot of people using it do. They have no respect for the people they stole art from to prompt and train it, they sometimes pass their generated art as their hardwork, painted and crafted themselves, and I believe that sometimes, cutting corners and going to the shortest route isn't the best Idea. Sure, I love ai-driven softwares that help me by reducing the time I spend on grunt work, like the generated trees you were talking about. However, I also like to fine tune and fine detail my work, as I believe it will give better outcomes for me and the people I work with. I also believe that AI art cannot come up with something new, it just regurgitates the things it was fed, and I know I just KNOW that if we rely too much on ai to save time and money, inovating and creating new visuals will just be rarer and rarer in the industry, resulting in games that will pretty much look nothing new. We already are neck deep into the AAA industry seeing games as a product that people consume, with some kind of formula to replicate and fine tune, rather than a medium that needs room to express itself and breath easily between innovation and inspiration, with enough time and money to mature. I don't believe AI will improve that side of the game developping industry. So, I'm not shaking in my boots like I first was at the begining of it, as I don't think AI will completely kill my job now, but I am weary of big wigs heads of studios seeing this as a way to Taylor their way into game design. Ai should never be a replacement for artists, engineers or writers, but rather a tool that should improve pur working conditions.
The fact that generative AI works by replicating patterns is what makes it so tempting but so bad. It's Uncanny, which makes it great at tricking people(esp. tech CEO's) but actually quite useless at making content that's meaningful or interesting.
Hey AI and games great vid as always. Not sure if you covered it already but inworld api seems very interesting AI tool. I come across a dev (vexfx) using it for a 3d character (within unreal) that his twitch chat can interact with. As the API allows you to dial in the persona of a AI. here is a quick link of what I recorded from his early work with it. Hes currently creating more rooms and inputs for the character to interact with. ruclips.net/video/qNrgVAXD66M/видео.html
Imagine living in a world where consent is required to copy the style of someone. Anime? Goodbye. Every Pixar clone made by not-Pixar? Banned. Learning by reference in art class? Not allowed. Let's hope the courts rule in favor of sanity.
What I find interesting the sheer vitriol towards A.I generation on the internet as a whole. This is the same internet that says "It's OK to use this intellectual property without permission to make a game! I'm a fan, of course it's fine!" and "Well, I wasn't going to buy it anyway, and I disagree with one of the people involved in the production, so it's okay to pirate it!". The incorrect assumption that A.I image generation is simply making a collage from a library of stolen images would go a long way to explaining things, but there seems to be more to it than just that.
@@AIandGames The words and phrasing used in this intro is much similar to the people that hold AI up as a great new thing. We can not frame generative “art” as anything but an algorithmic amalgamation of choices and art we’ve made now being shoved together to answer a prompt. Not only is it a gross misunderstanding of how we make art in this world and everything that influences it but also it trivializes the human experience to something that can be simply made and mass produced by a system that understands none of it. Art is beautiful because of its complexity and simplicity mixing together the emotional and technical meeting together to translate intent and experience through a process that is unique to each and every one of us. Even the art of a child with little to no technical knowledge or practice can still be just as beautiful a translation. Art can never be generated, never truly. The living aspect is simply inherent and can not be created by anything that doesn’t also live.
Thanks for watching this episode of Artifacts. If you can spare 10 minutes, please share your thoughts on generative AI for games in my viewer survey. We'll be coming back to your thoughts on the subject in a future episode.
forms.gle/1KmuEf3Qkv7y3yPy8
herez a problem generative ai could solve for 3d artists that wouldnt upset anyone: uv mapping. if i could ask a tool to uv map my models problem-free i would do it 100% of the time. uv mapping isnt fun, it isnt creative, its just tedious nothing work.
Even if it is capable, I absolutely do not trust studios to invest the proper time and effort required to make sure its output is good and not just used as a means to cut cost and not hire proper writers/artists. But all too likely, that's the direction it'll go due to the profit driven corporate heads at game studios.
This video is great, it covers all of the nuanced points I had while watching the video.
Thank you for covering this stuff in an actual lucid way, not just going "AI is the future, yeehaw!", and thank you for specifically addressing the legal issues and the capitalist issues of studios just wanting to increase their profit margins by laying off real people.
Also thank you for not acting like twitter artbros advocating for stronger IP laws which will only help companies like Disney
@@Erebus2075 OK. AI will never be able to do those things in any meaningful way. Do you even know how intensive modeling is? Animation? Writing Dialogue? AI will never get good enough to do these things because they're so intricate, so multidimensional. I know where you're coming from, but don't fall for the hype, you're just gonna be left holding the crypto bag or owning worthless property in the metaverse.
Great video! Not falling into AI hype and actually explaining the situation.
I wrote it in the Survey and I will say it in the comments. Everywhere were a Generative AI would generate content "on the fly" QA will be an issue and if we know one thing for sure. People will try to get any AI to say/do/request the most problematic thing possible
The lack of reproducibility, combined with the immense challenges of localization & cultural-appropriateness (a moving target across markets/time), just makes it all aspirational at best, and a fool's errand at worst. All the challenges and issues of PCG, only sold with the idea that you can ditch the humans who'd have been able to fix/address those issues.
@@mandisaw that's also an issue. I am not too familiar with the challenges of localisation to give an educated opinion
How energy efficient is the AI going to be? A topic no one talks about, I don't know why.
Small LLMs can be ran on a moderately powerful CPU with ~16 GB of RAM, if the dialogue were generated on a server and some smart dialogue reusing was implemented (eg. Generate dialogue response for 1000 players instead of unique for each player) it could be rather power efficient for it's purpose.
As for image generation we have Stable Diffusion that can generate high quality images in seconds on high end GPUs, again, if we smartly reuse assets we could reduce power consumption to be only a fraction of what the game consumes just for traditional rendering.
As for music, it takes less than a minute to generate a 30 second clip.
Overall, if the generators are implemented in a smart way they can be rather power efficient.
And a lot more power efficient compared to RTGI/ray tracing in games, or just the game itself being unoptimized.
There's also a lot of space for optimization, SD is already able to run on an IPhone thanks to some massive optimizations, so power consumption and GPU performance demand can be reduced further.
@@nocturne6320 Well, stable diffusion is pure plagiarism. For a good artist with a good vision for the end result, stable diffusion is simply not going to be good enough as a tool, ever.(art printer for easy money, more likely)
Back on topic, I am talking about cloud based AI with large tech companies behind them. I don't think small scale AI is going to go anywhere except for some really niche and specialized cases.
I suspect that the processing power needed for AI is going to increase exponentially to expanding size of dataset. I think the companies like OpenAI is cleverly crafting the dataset and constraining the AI to make it seem like everything is efficient and once people get reliant on it then they will open up the flood gates.
I'm saying it is efficient "now" but there are no real official data about this only some speculations floating around. So, there is a good chance it is energy inefficient compared to... -well, humans.
It's going to be like Bitcoin, but even better!
It's a next level search engine that doesn't just provide websites related to the question. But answers based on the question and information that is on the internet.
Great Video! As an engineer, I struggle with the "friendly" face of AI. I find myself typing "please" into my ChatGPT prompts because I keep lapsing into the default way I would talk to a human. I ended up spending a lot of time studying the transformer model, and understanding the "how" did help me use it more effectively to some degree (I can now put "prompt engineer" on my resume 🤮)
I now have a bunch of tips and tricks for using it well, but I would greatly prefer if the system had a clean, technical API with a manual, fixed rules for the input, and a consistent, repeatable output without variance. My current workflow feels like a never-ending code review with a distracted intern.
Oh boy... I really hate the term "prompt engineering" 😂
I do have a 'How does GPT work' video planned for AI 101 at some point.
It's important to say please to ChatGPT.
When the robot uprising happens it ensures your politeness and respectful behaviour is in its training data, which leads our robot overlords to show mercy on you and maybe the rest of humanity.
@@PlebNC RandoAI in 2040: "Brevity is the essence of wit. Stop apologizing with every paragraph!"
Fantastic video! This has got me wondering what the legal ramifications are if you own all the training set and generate new assets with it. Do you own it because it's based off your set or because it was made by AI would it be open? 🤔
Can you make more videos like this on AI more specifically on the future of npcs when integrated with Chatgpt.
I see AI artifacts at most as something being a cheap replacement when you don't have the budget to make it properly, but never will it be able to replace quality.
Great video! Confirmed a lot of my thoughts about the topic. Although I wouldn't expect companies to not use AI to replace humans. Of course they will do that. Two big companies both heavily invested in AI technologies have both fired thousands of employees recently and I would assume this is only the beginning of more to come. I am a solo games developer but I do not use AI technologies that scrape other artists work for obvious reasons (a lot of why mentioned in this video). Although, I wouldn't be opposed to making my own AI that scraped my own work to proceed to generate from. I think that would be an ethical solution to a lot of the legal problems for a start. Although, that might require a pretty large data set (maybe) but you could technically generate work from your own creations and then use your skills to fix what is produced and then feed it back into the training data and so on. I would probably attempt something like that.
I mean dev's have already been creating algorithms to generate levels etc in procedurally generated video games before generative AI became a buzz word for the media. Although one can assume it was done a lot more ethically e.g. creating the rulesets for how to generate the levels rather than copying from others work etc.
I am genuinely curious to see how these tools will evolve
For now we assume they will keep the same input forever, type a sentence -> BOOM you got a game
But, when you talked about speed tree I thought, "in the end, people will want many more handles to tweak the output, ensure it is consistent and reproducible, and be able to anticipate what will come out of it"
So, in my opinion, these systems will ultimately be getting more detailed, with dedicated interfaces, more precise and technical handles.
And, ultimately, might lose the source of the current hype
Not sure what I'm saying makes sense, morning coffee didn't kick in yet
That's essentially what happened with Stable Diffusion.
You can generate images just with text, but only when you fully utilize the other tools (controlNet, inpainting, regional prompts, etc.) you truly unlock the full power of the software. Giving you full composition over the posing of characters, composition, etc.
In my humble opinion, Generative AI in 5 years is gonna be so much more interesting than it is now. But it's probably gonna be a lot less interesting to many of the people currently advocating for it. Which I'm largely okay with
when will there be an ai adventure RPG game with actual working rpg mechanics?
I just finished the survey but remembered a crucial aspect: Most of these generative AI applications are not tailored to support the human creative process. Generative does not equal co-creative, and even some co-creative stuff does not address the creative process in the best way. Which is the best way? Ask the people you're trying to support with your system!
Amazing video, as always.
You just don't know how to use them co-creatively
@@nocturne6320 How is that even an argument? Do you believe Generative = Co-creative? Do you have examples in which the developers especifically intended to make their generative application to support the human creative process? furthermore, do you think that the fact that an application is co-creative makes it instantly and seamlessly fit into the creative process of the designers/creatives it's trying to support? All that information could help further this discussion.
nah I just can’t wait so I can create my own game with AI
Sure, we may need a person in the mix... Or at least there will be people that wanna use AI to help develop all kinds of things... The question is though, do we need more than one person. At a certain point, does any game creator with a firm idea what they want done need a single other person to get his vision realized. Regardless of how grand that vision is. Sure, right now there are alot of limitations... But what about next year? Or 2 years from now? 5 years, 10 years or 20 years? When do we reach the point where we can tell an AI "make the Matrix" and it just straight up does? And at that point, why would we need a second developer? And who's to say an AI can't formulate its own vision and create something all by itself. Overbelief in human exceptionalism is likely about to be difficult to hold onto over time.
When the customer knows how to perfectly define their requirements and define it to the AI program. We are out of job and since that is not going to happen, rest assure we will not be replaced.
That is until skynet comes for all of us.
> AI needs to be useful to game makers
Games AI researchers disagree with you here, they think that it just needs to be *interesting*
Ah, another video from That Scottish Man. Marvelous
I like to think this is how RUclips announces my videos to you all.
AMAZING VIDEO ON AI THIS WAS A MUST TO COVER
This video seems to be more a personal commentary than an educational one about the technology. Maybe not new, maybe hype. But still would want to learn more about it.
Anyone that believes that AI can do any art that we can lives in a truly sad world where they think that the art we create is not unique or special but just something that can happen. Even if someone handed you a piece of art that you somehow couldn’t tell was generated (unlikely) the second they tell you it was generated by a computer it loses all meaning and weight because instead of an expression of human experience and effort it’s just a lifeless emotionless generation that didn’t know what it was doing and just happened to do something that looks decent.
The WORST is people talking about how we can use AI to communicate with each other, writing all our messages for us...
If you're going to send a letter to your child congratulating them on a milestone, the least you can do is put in the minimal effort and time to compose it.
I would HATE to grow up in a world where parents are even less attentive than they already are
Counter point. The generative inputs and the construction of gans or other systems for AI generation represent both the "human" element or intent as well as being in themselves example of art. And to deny art in ai is to deny art as process or art in intent. Thus your stance is equally sad as it attempts to restrict the scope of what kne considers art or the weight of meaning.
A sculpture made using old techniques does not supercede one made using newer skills or tools toolsets in artfulness.
So it's sad you can't appreciate the art of development of these tools and their intelligent use cases as being unique or special when these toolsets do rely on the interactive human element to shape their outputs and thus in the right use case are equally valid as being thoughtful human art.
Much like an audio book or the use of dictaphone or speech to text technologies in writing have no bearing on the quality of a novel or peice of writing.
Further you seem to insinuate that "human" art being "good" has a prerequisite of the artist "knowing" what they are doing. Dismissing the countless instances of mistakes, errors, lack of technical skill or random impulse being core to the creation and mystique of art.
@@Ashamedofmypast This just proves my point nothing you said here makes any sense. The magic inherent to the art a living being makes can not be replicated by a program shoving things together trying to replicate what we make based on an algorithm of choices
@@clementineshetheyfae8312 everything I have stated is salient (typos aside) and in no way proves your point.
However both your comments show your complete lack of understanding of what these tools actually do and how they go about doing it.
And by extension suggest you did not actually watch the video. (Funny)
The fact you fail to grasp the art of construction/the art of application speaks volumes l.
You have opted to create a restricted definition of what the usecase and function of the tools are. Subsequently basing our opinion on the erroneous logic born from that definition.
Additionally you have failed to articulate how I am mistaken.
Rather than doing so you hand waved and dismissed my critique, offering no substantive counter argument.
Before reiterating your initial assertion.
Perhaps you should actually take the opportunity to understand the concept of AI generation, its applications, its intricacies in relation to the operator and the artful construction of the systems themselves.
As it stands the argument you make of the human element is a paradoxical self refuting strawman. Quite an impressive amount of logical fallacies to produce with such little text I must admit.
There's plenty of AI art that I've seen that I think is beautiful. Just like one might consider a natural landscape shaped by random processes and events to be beautiful, AI art can achieve the same. Not that I want AI art to replace traditional art, because traditional art still blows AI art out of the water by far, but to say that AI art is never beautiful is a bit much.
Hype?
Vegeta: I AM THE HYPE
32
We are among game connoisseurs and we only play highly crafted creative stuff right ? Ahrem....
The argument of quality for me doesn't really cut it as we've seen time and again on most creative industry the dynamic is lowering the creativity part to the minimum needed and just copy paste whatever previous thing work well, but producing it cheaper.
In the race to the bottom AI will be king, by lowering the barrier of entry to generate low quality games or content it will spam stuff until one game, for some bizarre random reason, is successful and floods the market.
Most commercial ventures prefer repeatable low effort / low return compared to a once in a lifetime high risk / high return.
Any tool that makes developers work easier also replaces developers. I mean, when that tree generator appeared, what happened to all the artists that were in the studio to create those trees? There probably wasn't a group of artists dedicated exclusively to creating trees but, if like 20% of time was used doing that then once that tool is in play 20% of those artists could just be fired without reprocutions to the time it takes to create a game while also making it cheaper to produce.
Replacemente doesn't come from ai doing people jobs, it come from a person using ai to do the job of multiple people.
Well there is an element of the scale and ambition of products growing as the tools do as well. Earlier games didn't really do natural environments all that much, so tree art jobs didn't really exist to be replaced in the first place. That being said the scale of this new wave of AI automation, or at least the promise it is a whole other beast.
"Are you utilizing any generative AI tools as part of your current role?"
This question should be improved. No, I'm not using generative tools at the moment, But not because of my discomfort or the studio policy. There are millions of different production-related reasons why are not using it at this stage. So there is not an answer that would allow me to answer properly.
That's fine. Providing the extra context as to why you are not is useful.
As a 2d/3d artist, AI art in itself doesn't bother me, but a lot of people using it do. They have no respect for the people they stole art from to prompt and train it, they sometimes pass their generated art as their hardwork, painted and crafted themselves, and I believe that sometimes, cutting corners and going to the shortest route isn't the best Idea. Sure, I love ai-driven softwares that help me by reducing the time I spend on grunt work, like the generated trees you were talking about. However, I also like to fine tune and fine detail my work, as I believe it will give better outcomes for me and the people I work with.
I also believe that AI art cannot come up with something new, it just regurgitates the things it was fed, and I know I just KNOW that if we rely too much on ai to save time and money, inovating and creating new visuals will just be rarer and rarer in the industry, resulting in games that will pretty much look nothing new.
We already are neck deep into the AAA industry seeing games as a product that people consume, with some kind of formula to replicate and fine tune, rather than a medium that needs room to express itself and breath easily between innovation and inspiration, with enough time and money to mature. I don't believe AI will improve that side of the game developping industry.
So, I'm not shaking in my boots like I first was at the begining of it, as I don't think AI will completely kill my job now, but I am weary of big wigs heads of studios seeing this as a way to Taylor their way into game design. Ai should never be a replacement for artists, engineers or writers, but rather a tool that should improve pur working conditions.
intellectual property is bs.
nah I am excited to have AI come in and generate a game that I want
i just hope ai doesnt start replcing 3d environment artists before i get a job doi g what i love lmao
The fact that generative AI works by replicating patterns is what makes it so tempting but so bad. It's Uncanny, which makes it great at tricking people(esp. tech CEO's) but actually quite useless at making content that's meaningful or interesting.
Hey AI and games great vid as always. Not sure if you covered it already but inworld api seems very interesting AI tool. I come across a dev (vexfx) using it for a 3d character (within unreal) that his twitch chat can interact with. As the API allows you to dial in the persona of a AI. here is a quick link of what I recorded from his early work with it. Hes currently creating more rooms and inputs for the character to interact with.
ruclips.net/video/qNrgVAXD66M/видео.html
Yeah I highlight Inworld throughout the video but we''ll be looking at their tech in more detail in a future video (very soon in fact).
@@AIandGamesnice, looking forward to it.
AI gonna make my dream come true in the future cuz I will be able to generate games with AI
Imagine living in a world where consent is required to copy the style of someone. Anime? Goodbye. Every Pixar clone made by not-Pixar? Banned. Learning by reference in art class? Not allowed. Let's hope the courts rule in favor of sanity.
What I find interesting the sheer vitriol towards A.I generation on the internet as a whole. This is the same internet that says "It's OK to use this intellectual property without permission to make a game! I'm a fan, of course it's fine!" and "Well, I wasn't going to buy it anyway, and I disagree with one of the people involved in the production, so it's okay to pirate it!". The incorrect assumption that A.I image generation is simply making a collage from a library of stolen images would go a long way to explaining things, but there seems to be more to it than just that.
Complex copycats. It feels like this description also applies to humans 😅
19:48 oh here we go...
Time to burn down the copyright system to its core... There, that solves everything!
No.
Moving on.
This intro is not reassuring
What, you don't wanna use Bing?
@@AIandGames The words and phrasing used in this intro is much similar to the people that hold AI up as a great new thing. We can not frame generative “art” as anything but an algorithmic amalgamation of choices and art we’ve made now being shoved together to answer a prompt. Not only is it a gross misunderstanding of how we make art in this world and everything that influences it but also it trivializes the human experience to something that can be simply made and mass produced by a system that understands none of it. Art is beautiful because of its complexity and simplicity mixing together the emotional and technical meeting together to translate intent and experience through a process that is unique to each and every one of us. Even the art of a child with little to no technical knowledge or practice can still be just as beautiful a translation. Art can never be generated, never truly. The living aspect is simply inherent and can not be created by anything that doesn’t also live.