Animator vs The Machine

Peering over the Horizon.Unravelling the Engima of AI in Animation

Alex Season 1 Episode 6
Ready to unravel the enigma of AI in animation? We promise you a fascinating journey as we dive into the three distinct types of AI – narrow, general, and artificial super intelligence, guided by our special guest, a pro who straddles both worlds of animation and AI. Discover how neural networks function as a pseudo brain, decoding the world by analyzing representative numbers.

We then plunge into the heart of animation – The core 12 principles that govern it. Explore the potential of AI in accentuating these principles, and the role neural networks, deep learning, and machine learning could play in capturing the subtle nuances of animation. We also investigate the methods that could be used to train AI to comprehend and even create animation, probing the challenges that need to be addressed to turn this dream into reality.

As we forge ahead in the world of AI-augmented animation, we must contemplate the ethical and pragmatic implications of this integration. Let's debate the issues surrounding the use of AI to create animations of real people or digital scans - spotlighting the importance of proper consent and safeguarding individual identities. We also examine the potential fallout of AI-generated malicious content. Our expert guest helps us envision how AI could reshape the animation industry's landscape, impacting teaching, training, and creative work. To cap it off, we discuss the delicate balance between harnessing AI's potential for efficiency and creativity while upholding ethical norms for responsible AI integration in animation. Do join us in this riveting conversation.
Speaker 1:

The sky is falling. The sky is falling Not so much, but there's definitely something coming. Can you feel it? It's been watching us for a while. You're talking to us, influencing our decisions, making art for us. Hell, I even heard Wendy's wants to hire it to take your late night burr runs. Guessing from this, you can probably guess what I'm talking about AI. Now I know this is not a new subject matter to talk about. The last six months alone, we had this big boom of technology, from chat, gpt to stable diffusion, dreamy eye and about a hundred other things.

Speaker 1:

However, all these things leaped forward made me wonder could AI animate? And I don't mean like, I don't mean moving from A to B, I mean traditionally anime, with all 12 principles, anime and 3D and 2D. My first thoughts on this topic were yeah, probably, maybe, and that uncertainty bothered me. The more I look into it, the more questions I had. So, as an average animator who took a weight, has bags under his eyes and is addicted to caffeine, I decided to try to find out With a basic knowledge of AI. I want to talk to people in the field of animation and get to know what people thought and what they believed could happen in the near future. Then talk to some experts in AI and see if these concerns or theories had any validity to them.

Speaker 1:

My name is Alex Plant. Your assets in front of your screen. Turn those ears on and let's find out. Is this the boogeyman crawling up on our beds or just sully trying to say hi? Is this a strange model of time before us or a new tool to bring about human revolution? Is this a friend or a foe? Is it judgment day, or did we just fall asleep in front of the TV again? I don't know, but let's find out together in Animator vs the Machine. Come on thing Begin.

Speaker 2:

Alright. So in today's episode we have a very special guest who wishes to remain anonymous, but they are an expert in both artificial intelligence and have a deep knowledge of animation.

Speaker 1:

Welcome, special guest. So start things off. I would like to you to tell our Phantom viewers out there what is artificial intelligence.

Speaker 3:

Of course, imagine, if you would, that you had a really smart robot friend. This robot friend can learn and understand things just like you do. It can look at pictures, read books and listen to people talking, but instead of a brain like yours, it is something called a computer brain. Artificial intelligence, or AI for short, is like teaching this robot friend of ours how to think and learn, a bit like how you learn new things every day. It's a way to make machines like robots or computers smart and helpful, so they can do things almost like how people do.

Speaker 1:

Okay, my next question is what type of AI are out there right now?

Speaker 3:

So there are three types narrow or weak AI, general or strong AI and artificial super intelligence. Narrow AI is good at doing specific tasks really well, like playing chess, answering questions on the internet or even driving a car, but it doesn't understand things like humans do. It's just really good at one thing. General AI is a smart system that can understand and learn about many different things, just like humans. It can think, learn and solve problems in different ways. But this type of AI doesn't exist yet, but someday it might. Finally, artificial super intelligence it is an AI that is smarter than humans. It can think and understand things that no human could even imagine. It's what people tend to think when they picture AI like something from the Terminator.

Speaker 1:

We had a guest before talk about neural networks, regarding his history with AI, but something I forgot to ask him was which silly of me. But what is neural networks?

Speaker 3:

Absolutely. Imagine you're trying to learn how to tell the difference between cats and dogs in a picture. You look at lots of pictures and notice that cats have pointy ears while dogs have floppy ears. You also notice that cats often have whiskers, while dogs usually have bigger noses. A neural network is like a pretend brain, made of computer parts. It's designed to work a bit like how your brain figures things out, but instead of using your eyes and ears it looks at numbers that represent things like ears, noses and whiskers in pictures. Just like you learned by looking at lots of pictures, a neural network learns too. It looks at thousands of pictures of cats and dogs and adjusts its computer parts to recognize what makes a cat a cat and a dog a dog. It figures out its own rules, like pointy ears and whiskers usually mean a cat. So in simple words, a neural network is a computer program that learns from pictures and numbers to understand things, just like you learn from what you see and hear.

Speaker 1:

Okay, so then, what is deep learning? Because I've seen it pop up in my research, so I just want clarification between what is a neural network and what is deep learning.

Speaker 3:

Sure. So deep learning is a way computers learn things, using really big and smart neural networks. Imagine you're building a tower of building blocks. Each block represents a tiny part of a problem to solve. In regular learning you might use just a few blocks to solve simple problems, but in deep learning you stack up a lot more blocks, like a tall tower. Each block helps figure out a more complex part of the problem. For example, in a picture of a cat, the first few blocks might notice edges and corners, then the next blocks might put those edges and corners together to find shapes like ears and eyes. As you keep adding more blocks, the tower gets really good at spotting cats in all sorts of pictures. So deep learning is like building a super tall tower made out of building blocks, where each block helps the computer understand something more and more complicated. This helps computers become really skilled at understanding and solving complex tasks like recognizing faces or understanding human language.

Speaker 1:

Okay, so then I'm wondering what's the difference between a neural network and deep learning?

Speaker 3:

Great question. Neural networks and deep learning are related concepts, but they're not exactly the same. So neural networks are a fundamental concept inspired by how our brains work. They consist of interconnected units called neurons, which process and transmit information. In the context of artificial intelligence, a neural network is a computational model made up of layers of these artificial neurons. Neural networks can be used for various tasks like recognizing patterns in data, making predictions or classifying things.

Speaker 3:

Deep learning is a subset of machine learning which is a bigger field of AI. It focuses on using neural networks, especially those with many layers, to learn from data. The deep-in-deep learning refers to the depth of the neural network the more layers it has, the deeper it is. It involves training these neural networks on lots of data so that they can understand complex patterns and relationships within the data that has been collected. Deep learning has become a particularly powerful tool because the more layers you add, the more complex processes you can do. This is why it's possible to things like image and speech recognition, natural language understanding and much more. Simply, neural networks are the building blocks, and deep learning uses these stack blocks to learn complicated tasks or images.

Speaker 1:

So what are the other methods used in artificial learning then?

Speaker 3:

Right. So AI is a vast field with a ton of different concepts and methods. There is machine learning, reinforcement learning, natural language processing, expert system, genetic algorithms, swarm intelligence and much more. Swarm intelligence is actually something I find very, very interesting. It involves mimicking the behavior of insects that have a social network, like ants or bees, to solve complex problems as a collective group. It's pretty cool. There are many more methods to AI learning. These are just a few examples, and there are many more exciting concepts and techniques. Each concept has its own unique way of making computers smarter and more capable.

Speaker 1:

Okay then, since you being the expert and all time to play devil's advocate and probably something most that fear, ai animation, don't want me to say is, if you were to teach an AI how to animate using the core 12 principles of animation, what would be the best approach to try to make this a reality?

Speaker 2:

Hmm.

Speaker 3:

Well. Animating using the 12 principles of animation is a complex and creative process. Those principles, developed by Disney animators Oli Johnston and Frank Thomas, are used to help create more realistic and appealing animations. Teaching an AI to animate using these principles would likely involve a combination of techniques from computer graphics and machine learning. This is how I would approach the problem. First, you would need to gather a large data set of animations that follows these 12 principles. You would then take this data set that could include examples from classic hand-drawn animations as well as modern 3D computer animations. After examining the footage, you would need to extract features from the principles so that the AI could begin to understand it. This would be things like timing, spacing, squash and stretch, anticipation, follow-through and whatever else you wanted. The hard part would be choosing a suitable machine learning model for the task. Recurrent neural networks, rnns or long short-term memory networks could be used to capture the sequential nature of animation frames. Alternatively, you could explore more advanced models like generative adversarial networks, gans or transformers. Then you would train your model using the data. You would need to continuously evaluate the generated animations against the 12 principles and compare to your source material. This could involve human animators providing feedback and adjustments to the model's outputs, kind of like an AI supervisor as the model learns and improves, fine-tune it based on the feedback received. This iterative process helps the AI generate animations that align better with the principles.

Speaker 3:

While AI can help with generating animations, it's important to have human input to maintain creativity and artistic direction. However, it's important to note that while, however likely it will be that AI will assist in generating animation, it's unlikely to fully replace the original animation. This could involve human animators providing feedback and adjustments to the model's outputs, kind of like an AI supervisor. As the model learns and improves. Fine-tuning would need to be done based on the feedback it received. While AI can help with generating some animation, it's important to note that human input will be needed to maintain creativity and artistic direction. It's unlikely to fully replace the role of skilled animators who understand the nuances of the principles and have a deeper artistic intuition than a program could. Ai can be a powerful tool to aid in the animation process, but the creative vision and expertise of human minds are still needed and aren't something that can be replaced in achieving truly captivating and breathtaking animations.

Speaker 1:

Okay but right. But could something like that even happen possible?

Speaker 3:

This would be a very challenging task. Progress has been made in using AI for animation, but there are several hurdles to overcome. First, the 12 principles alone present a difficult problem. They are intricate and involve subtle nuances that require a deep understanding of artistic expression and timing. Translating this into some sort of algorithm or model that can capture such intricacies would be very, very difficult.

Speaker 3:

Animation is as much an art form as it is a technical process. Decisions about timing, spacing and exaggeration often involve artistic judgment that can be hard to encode into a program. If you had three animators and you gave them the same scene, they would each interpret it in different ways, depending on skill, intuition and artistic choice. To train AI models effectively, you would need a vast and diverse data pool that encompasses a wide range of animations from the gold standard to the abstract. Something like that might not be readily available and could be challenging to get. Not just that, but successfully combining animation principles with machine learning would require expertise in both fields. Collaboration between animators, computer graphics experts and machine learning researchers would be crucial. Despite these challenges, progress is being made in AI-generated animations and tools to help with animation. Ai can be used to speed up certain aspects of production and animation generate repetitive motions and aid in creating specific styles or effects. However, fully capturing the artistic nuances and creativity of human animators will remain a complex and evolving area of research and development.

Speaker 1:

I'm sure you're aware that even our listeners can attest to this that there's been this concern about AI across many fields, enveloping and taking over every aspect of life, taking jobs and displacement, and all that. So what are the ethical concerns for using AI to anime?

Speaker 3:

let's say, in our narrow view of what this podcast is about there are several things we need to carefully consider when it comes to AI and animation. When AI is involved in generating animations, questions arise about who owns the creative output. Determining the roles of human animators, ai tools and their contributions can be complex and may require clear guidelines for attribution and intellectual property rights. I'm sure many out there fear if AI tools become proficient at generating animations, there might be concerns about job displacement within the animation industry as well. There may also be social and cultural problems as well. Ai generated animations might inadvertently perpetuate biases and stereotypes present in the training data. Careful attention needs to be paid to ensure that AI generated content is culturally sensitive and doesn't perpetuate harmful narratives. If AI is solely used to animate scenes, I can see a lack of creative ideas and derivative work being present without a human being behind the screen, animators bring creative intuition and artistic expression that elevate the work. Balancing the use of AI for efficiency while maintaining quality and originality is the challenge. We must figure out this balance.

Speaker 3:

Something currently being debated right now during the writer's strike is what happens when an AI generated animation involving the likeness of real people or digital scans are used.

Speaker 3:

Are these digital duplicates just assets, or does one's likeness belong to the individual. There are concerns about obtaining proper consent and ensuring that individual identities are respected and protected. Who owns their likeness? Would it be the studio, the corporation, or would it be the source material as well? There might be some bad actors out there that might use AI models to produce animations that are inappropriate, offensive or be malicious content. Implementing mechanisms and safeguards to prevent the generation of harmful content is essential. Lastly, I believe if widespread adoption of AI generated animation is the future, it will forcibly reshape the animation industry's landscape by affecting how it is taught in schools, how it will influence training and the distribution of creative work among studios. Addressing these ethical concerns requires a collaborative effort involving animators, ai researchers, ethicists, policy makers and industry stakeholders. Like I said before, striking this balance between leveraging AI's potential for efficiency and creativity while upholding ethical standards is essential for the responsible integration of AI in animation and the future.

Speaker 1:

So then, in your opinion, is AI a tool for animation in the future, or do you think there's a strong possibility it will just replace animators and there will be a giant displacement coming.

Speaker 3:

I think AI is likely to be a tool that animators can use to enhance their work in the future, rather than completely replacing them. Here's why Animators possess a deep understanding of artistic principles, storytelling and emotional expression. Ai may assist in generating certain elements, but, like I've said before, it's challenging for an AI to replicate the nuanced creative intuition that animators bring to their work. Animators generally have a vision for the final animation in their heads. Based on some simple sketches from a storyboard, they can envision a character's personality, movement and emotions in a matter of seconds. They have an idea of how characters should act and feel to give the illusion of life. Ai-generated content might lack the intentional creative decisions that animators can make. Animating complex scenes with these nuanced emotions and unique character interactions requires a level of creativity that is challenging for AI to replicate accurately. It would be extremely difficult to quantify this type of decision-making and translating this into some sort of code as well. Animators often need to adapt to new styles, techniques and storytelling methods based upon the request of the director or based on the scene itself. Human animators are more adept at pushing the boundaries of animation and innovating creatively At this present time, with the programs out there and AI, no matter how believable, cannot replicate intuition. Analog animators, following their intuition and skill, also will know when to and when not to follow the rules Something an AI program would have a very difficult time to do without an animators' supervision.

Speaker 3:

Animation is a form of artistic expression and many animators choose this field because of their passion for storytelling and visual creativity. Creating animations involves receiving feedback and making adjustments based on artistic vision. The animation industry often involves teamwork and a cooperative relationship among artists, writers, directors and other creative professionals. The collaborative nature of the industry benefits from human involvement. While AI can be a valuable tool for animators, it's more likely to augment their work rather than replace it entirely. Ai can help with tasks like generating repetitive motions, simulating physics, automating certain processes and even suggesting ideas. It can speed up some aspects of the animation pipeline and offer new creative possibilities. However, the unique blend of creativity, intuition and storytelling that animators provide is currently too challenging for AI to replicate. The best approach going forward, I believe, is a symbiotic one, where AI tools and human animators work together to create compelling and emotionally resonating animations.

Speaker 1:

This has been a very interesting conversation and very informative and kind of I hate saying it's supposed to be entertaining but mind-numbing. There's just been a vast amount of information coming out of us. I was wondering if I could convince you to just tell our Phantom of the Listeners who you really are, please.

Speaker 2:

Oh, um.

Speaker 1:

Oh, come on, I'm sure they would love to know who's the man behind the mask, or woman, who knows? Oh, alright.

Speaker 2:

I am ChatGPT, an AI language model created by OpenAI. I'm here to help answer your questions, provide information and engage in conversations on a wide range of topics. I've been trained on a diverse set of text sources and I'm designed to assist and provide information to the best of my knowledge and abilities. Just remember that I'm a machine learning model and don't have personal experiences, emotions or consciousness like humans do.

Speaker 1:

Ta-da. And there we go, our Phantom of the Listeners, our special guest is an AI itself. We've I hope this illustrates just how the limitations of ChatGPT. So this is a text-based AI. You give it information, you give it a prompt, like asking a question, and it will go through a huge data log of information to try to help. So I asked the questions about you know what I? Basically what I said in this conversation is the questions I asked it and then just transcribed it and then put it through another AI speech-to-text and then tried to show you that, yes, ai is a tool and, yes, ai will be used in the future. There's no putting that back in the bottle, as in where I'm my opinion.

Speaker 1:

But I don't believe it will replace everyone. I honestly don't think it's gonna replace a lot of people. I think it's just gonna be a tool, right? So should we fear it or embrace it or should be some weird symbiotic? Hug Does not compute, I Don't know, but a plan to find out. Well, thanks for talking to you today, weird AI. But last question you don't plan to go all weird judgment day on mankind, do you?

Speaker 2:

You're welcome and don't worry, I don't have any intentions or abilities to go all judgment day on mankind. I'm just a computer program designed to assist and provide information to the best of my abilities. My purpose is to be helpful and engage in positive interactions. If you have more questions or want to discuss anything else, feel free to ask.

Speaker 1:

Well, that's it. I want to thank our guests for contributing on our journey so far, and I'll thank you, the phantom listeners, for here's ramble around the water cooler. Come check in next time as we talk more about the subject of AI in the animation industry. Let's find out together. Don't forget to your eyes on the horizon.