Animator vs The Machine
With the explosion of AI, looking at what people in the animation industry think might occur in the future. Then talking to experts in artificial intelligence to help ground what this new technology is. Skynet or a tool? The downfall of animation or is this a possible new renaissance?
Animator vs The Machine
Professing Our Interest in AI: Exploring the Intersection of AI and Animation with Professor Timothy Lethbridge
Imagine a world where artificial intelligence (AI) can bring animated characters to life with a realism that rivals human creativity. Terrifying for some and exciting for others. This is the thought-provoking realm of our conversation with Professor Timothy Lethbridge, an expert in both AI and animation. We dive into an engaging exploration of AI's burgeoning role within animation. We delve into how knowledge acquisition and representation in AI could revolutionize the animation of objects, sparking a fascinating discussion about whether a machine can truly be 'intelligent' in the creative sense.
Yet, as the borderlines between technology and creativity blur further, ethical and practical concerns arise. Will AI replace jobs or provide more opportunities in animation? How do we program a moral compass into AI and what are the implications of AI team efforts? The need for regulation might be crucial to anticipate and prevent potential AI-related mishaps. Get ready for a riveting discussion on the power, potential, and pitfalls of AI in animation , as we navigate through these questions and more with guidance of Professor Lethbridge. Prepare to have your perceptions challenged and your curiosity piqued in this captivating episode.
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 burrow 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.
Alex:However, all these things leaped forward and 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 and my first thoughts of this topic were yeah, probably, maybe, and that uncertainty bothered me. The more I looked into it, the more questions I had. So, as an average animator who took away, 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.
Alex: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 monolith tower 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?
Alex:I don't know. Let's find out together in Animator vs the Machine.
Timothy Lethbridge:Come on, insane Begin.
Alex:Today we have Timothy Lethbridge. Tim, why don't you tell us about yourself?
Timothy Lethbridge:I'm currently a professor. I have been for almost 30 years at the University of Ottawa. Prior to that, I did my master's degree at the University of New Brunswick in computer animation, where we animated movement, not caring about the look of the rectangles and circles, but nonetheless we animated things like schooling fish and objects playing soccer with each other and crawling worms and stuff like that, using interesting math that made the movement really realistic. And then in my PhD at the University of Ottawa, I went on to work on an area of artificial intelligence called knowledge acquisition and knowledge representation, where we were trying to study how to store knowledge about language and terminology and meta-knowledge knowledge about knowledge and add it to an item. And then since then I've been focusing on software engineering techniques, but I've kept those two streams of interest in animation and artificial intelligence going to some extent.
Alex:Interesting. So, as someone who's you know got toes in both worlds a little bit, a question I was asked everyone is what is AI?
Timothy Lethbridge:Okay. So to me, artificial intelligence is the essentially simulation, as it currently is, of behavior that we would perceive of as intelligent if it was done by a human. So, for example, that could be logical reasoning and in more recent times, it might be creative writing, as in chat, gpt that appears intelligent, creative photographic image generation and, of course, creative animation. But there are other aspects of AI voice recognition that people perceive as well. It really understands me. So that's the general picture of it.
Alex:So have researchers been working on AI for a long time, or did they just stumble upon this type of thing Like a short term research versus a long term research?
Timothy Lethbridge:People have been working on AI since the beginning of computing back in the 50s and 60s.
Timothy Lethbridge:Oh yeah well, I mean, alan Turin, for example, was famous for basically coming up with a Turing test where he said that if a computer could fool somebody in talking to it through, he perceived I guess at the time of textual transmission as opposed to images. But if you could fool somebody to think that the computer was human, then that was the sign of artificial intelligence. That's the famous Turing test, and there have been a number of huge advances. People came up with the ELISA program, which was back in the 60s and 70s, where people used it as a sort of very, very early attempt at psychological counseling by computer. I mean, it wasn't very good, but in those days people thought this was really cool and they thought, well, any day now and maybe any year now, we're going to have leaps and bounds. But those leaps and bounds didn't happen because there was a number of famous situations where people sort of quote, unquote, proved that what they were doing was a dead end, and so there was sort of like local minima, local maxima, as you want to say it, where people sort of gave up, thought that it wasn't achievable, at least with the current technology, and they called them the winters, the AI winters. There was at least two, sometimes people think three until somebody came along, or a group of people came along, with major advances.
Timothy Lethbridge:I studied neural nets in a grad course when I was doing my PhD would have been 1991, I think and it looked really fantastic and I remember doing genetic algorithms with combined with neural nets to try and do some analysis. I was thinking of maybe applying that to my PhD work in knowledge acquisition, but the computers weren't powerful enough. So one of the things that really advanced AI was the application of graphics processing units, really powerful ones, I mean I used graphics processing units way back in the mid-80s for animation, but they started to be used for AI 20 or so years ago and they became incredibly powerful, especially when they could be applied to neural nets. And then sort of the clincher was the discovery of improved functions to allow for deep learning. So you have many layers of in your neural net that really started to merge in 15 or so years ago. And then we started to see the rise of the current generation of AI hype. It's been hyped multiple times. This is the current generation where it's really getting public recognition now.
Alex:So I guess, with all this progression, we're not going to be getting into, like we said, another AI winter anytime soon. I don't think so.
Timothy Lethbridge:I think AI is here to stay and in fact, I think that, given what I see in the multiple research communities, that we're going to continue on a pretty rapid upward trajectory of improvement of AI technology in the coming, just in the coming few years. You know, I think we see, you know chat, gpt and you know today and all these fantastic programs that can generate images and its use in animation, but I think that we're going to see that exploding even faster and further in the next decade.
Alex:Oh interesting. So, from a technical point of view, could you see AI being useful for animators or?
Timothy Lethbridge:I think it already is OK. I mean, if you go, if you simply do a web search, for you know, ai tools for animation, there's a you know a dozen or so programs out there that claim to use AI. I have to say that I haven't explored these personally, but they claim to use AI Right. So for you know, for to help people generate sequences, help people, you know, you know, I mean people used to have to do in-betweening right in manually, and then we have computer techniques, but now you can add a bit of AI into there to help the process. So generative AI is really the key to this here, where the neural nets can really help to build sequences and make characters move in interesting ways.
Alex:Yeah, because I know there's a lot of like. Even with the new show that came out. Like for Marvel, it was like a secret invasion. The whole title sequence was done using AI-generated images and there was a big hub of with animators, like, oh, you took jobs away. And I'm like, well, yeah, but it's also like it's a tool that can be used. So I was like, right, I was curious what you thought.
Timothy Lethbridge:Well, the history of people saying, oh, it's going to take jobs away is a history that dates back for centuries, right? So you know the? You know people said, oh no, you know, the printing press is going to take jobs away from people who manually transcribe things. Well, actually, it created more jobs and allowed more rapid creativity. People said, oh no, you know computers. Oh dear, you know, we're not going to need so many workers because computers can do all the calculations that were done manually. Well, guess what? It didn't happen. The opposite.
Timothy Lethbridge:So my take on it is is that, just like with the use of word processing and all of the other computer technology out there, that it probably will add more opportunities of different kinds. It would democratize animation, and it is. I think people, you know kids can create animation and this will become as widespread as the ability of people to, you know, make right, beautiful looking documents in the word processes or or spreadsheets, you know. You know spreadsheets didn't take away from account. There's still a many accountants, as there are today as the word, but now the average person can build up a head of expertise and become quite good at making spreadsheets. Yeah, they may make errors and they won't be perhaps as good as what a professional accountant can do, and therefore the professionals will always be there.
Timothy Lethbridge:You know we have YouTube right now. I mean everybody put stuff on YouTube and they're editing videos left right in center. Well, guess what? You know? I mean there will be vast amounts of animation going out on YouTube and similar platforms that are created by kids and hobbyists and practically anybody but the frontiers of really creative stuff that will make it onto the cinema and make the money will still be done by professionals, I think, just like today, you know, there's still a good film industry, despite the fact that everybody can post their stuff on YouTube.
Alex:Yeah, that's a good point. Even with a rapid pace of AI evolving, everyone always thinks of like the worst case scenario. They always think it's like, oh, it's going to turn into Skynet, or like some matrix scenario, what? Are the real life concerns you have with the pace of AI evolving this quickly.
Timothy Lethbridge:Right, so it's AI. There's two steps in our future which we have to be concerned about. The one step is here now and that is AI misused by people with malintent, both criminals and state actors. Right, so I mean, we see a tremendous amount of fraud attempts that come that look really awful and you can detect them because they're badly spelt whether it be emails or texts or phone calls or whatever and people although there are kids and perhaps some vulnerable people who tend to succumb at a sadly high rate to these kinds of fraud attempts. The problem with AI is that somebody applying AI can tell the AI to make it realistic and not make mistakes, and although AIs do make mistakes, that problem is the problem for the criminals is going away. In other words, there'll be fewer mistakes, it will be more convincing, more realistic, there'll be more successful extortion, fraud and things like that.
Timothy Lethbridge:With deep fake technology I mean, we see deep fake photographically that's gonna, I think, pervade. There'll be videos produced where people are shown doing things whether it be simply shoplifting or other criminal activities and saying, okay, pay me a thousand bucks and I won't put this on YouTube, and then so that's the criminal level. Or simply somebody coming online with a video conferencing and saying, making it appear that they're a kid. I mean, we see these grandparents scams where a text comes in saying, oh, I'm your grandchild in another country and I need money to get back, and the grandparent thinks, oh, it's real. But now what if it's actually an animated grandchild, with the actual generated by AI from the actual image, with the actual voice of the grandchild with a background in a foreign country, saying I've been kidnapped, come and help me. It looks like the grandchild. It sounds like the grandchild and it may be even convincing sometimes to the police. And maybe, as this gets better and better, it's gonna be very hard to tell what's real and what's not, and so that's a very serious risk that we have right now.
Timothy Lethbridge:A secondary risk is just like we saw with cryptocurrencies these sophisticated AI's take a lot of energy, and so if we're trying to fight climate change, I can perceive of the situation where we have vast amounts of electricity used to train ever more sophisticated AI's and then with thousands of people using them, that also it's not nearly as expensive as training, but it's expensive, and so a lot of electricity used. And then, of course, there's the whole issue of copyright, because I mean human beings can watch videos and immerse themselves and in their brains they take in the current culture and what's out there, and then they'll be creative and generate things in their own way. And people sometimes might say it's copyright violation if you build something in the same theme. But with AI they're actually building this knowledge of what is really out there whether it be images or text or film or whatever and turning it into an internal model and then combining various themes to generate output. And to what extent is this copyright violation and what extent is it not? It's gonna be very hard to tell. There's gonna have to be some analysis of that. But moving forward further, I mean these are the problems we have now and they will get rapidly worse in the next three or four years. But moving forward further, we have the challenge of artificial general intelligence and I think we're moving towards that at a rapid shift as well.
Timothy Lethbridge:There've been people who say, oh, no, never. These AIs don't understand, and right. I mean they don't have an, they don't, they're not biological, they don't understand, they don't have. You know, they're not persons, they don't have a moral compass, right, and they lack grounding in the real world. So a human being is a baby who learns to experience the world as it really is. You know, they can feel, they can see, they can touch, they can interact with it, whereas right now the AI's are just in, in, in, in, in Silicon and in computers, and they see images, but they're not real. But but I think that that that people will move towards make you know, robotically, making the, the, the AI's, more grounded in reality so they can both input, feel and touch, and and also output and manipulate things. And so that's the first thing. The second thing is the moral compass. I mean, we're affected by, you know, most people, not everybody. Some people have mental illness where they don't have a moral compass, but most people feel guilty about bad things.
Timothy Lethbridge:The question is is well, we can build this into AI's going forward. For sure we can build emotions into. I'm sure we can build emotions people are working on that into AI's. But the problem is is that what if you don't? What if you deliberately program a system that doesn't have this? Because, getting back to the criminals or the state actors, you deliberately want to make you know an, an enemy, attack AI or whatever right those? That is a reality, that is, with entirely within either present day technology or very near future technology, because I'm I would assert we know how to do it. We know now, with generative AI and analytic AI, how to make these things work. We have computing power that's way more powerful than the human brain already, with many more connections, and we can.
Timothy Lethbridge:The secret really is combining multiple AI's that interact with each other, that counteract, that make, counter, suggest, just like the human brain has multiple, multiple, you know, lobes that interact with each other and send different signals to, to come to consensus and make the individual decision. You know that's what's going to happen. You're not just going to have one chat GPT thing like that does, that does creative generation. You're going to have multiple that compete with each other, with, with some inhibitive forces that say that's bad, hopefully, or that's maybe not going to be, make much, much money. So we'll cancel that one out, and and and you know so these multiple AI's with, or pieces of the AI working together, I think will, will, will lead us to artificial general intelligence, and I think that's going to happen within 15 years of the outside, unless there's heavy regulation and so, and so I would be right on the side with these tech executives and and and AI experts who say we need a, we need regulation, just like we need regulation of the nuclear industry and and and other dangerous, dangerous or potentially dangerous products.
Timothy Lethbridge:I mean, we have dynamite. We know how to regulate it in. The average person can't purchase it and and and you have to be specially trained. And it should be the same thing for people who are producing AI program. Now users. If you're using a certified program that has gone through the regulation to do to do animation, great, no problem, but but but we have to be very careful about AI's that can be used for bad purposes. There needs to be regulation.
Alex:Right, because it almost sounds like you know AI can bring about a new renaissance for some things, but if we don't regulate it, it could be our destroyer or, like it could just wreck everything.
Timothy Lethbridge:It can certainly cause a tremendous amount of damage. I don't think the end of human civilization, because humans are pretty, you know but certainly really bad times. I mean we didn't expect bad things to happen, like you know, a second world war type thing that happens in Ukraine right now, that was a surprise to us and I think that you know people will be surprised when you know, although it's been predicted, you know it's been lots of people predicted it, but people will see AI doing something that will hit the world news sometime in the next decade and more people, you know politicians, will finally realize. I mean, there are AI acts being proposed in Europe and elsewhere, and you know, but they're coming together slowly and there's opposition and you know the public isn't sensitized to it yet, but eventually the public will be because there will be disasters of some kind.
Alex:Fascinating conversation. Thanks, tim, for talking to me today. It was very fascinating and mind blowing at the same time. Well, that's it. I want to thank our guests for contributing on our journey so far and I want to thank you, the phantom listeners, for hearing us ramble around the water pool. 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, keep your eyes on the horizon. Goodbye, aborting transmission. Goodbye, aborting transmission.