SORA AI: When Progress Is A Bad Thing

today's video is sponsored by conflict of Nations more from them in a bit I like making Art here's a picture I drew of a frog here's a picture I drew of a dog all of this is fun for me you know I guess it gives my life meaning which is why I don't like AI art I know some people don't like AI art because it steals from artists the robots are stealing the picture and they're smooshing them together or something like that I don't care about that see I'm more concerned about how this thing makes everything that I do completely irrelevant AI art threatens to do everything that I can do but faster my biggest fear in 2023 was that in the future starving Indian children would learn how to use chat GPT to write video essays you know they'd use the AIS to edit the videos together and they' used voice cloning tools to voice this stuff to make them sound more American with all of this technology they the enemy could threaten not only my hobbies but my livelihood so I decided to do something bold an experiment and it went a little something like this conflict of Nations is a free online PVP strategy game available right now on both PC and mobile platforms choose from a real country that you will lead during World War II fight up to 128 other players in realtime battles that can take weeks to complete while using a variety of different units to build your unique Army declare war with your neighbors or Forge alliances with other players for the time being make your own strategies test them out all while trying to eventually take over the world the game is available on both PC and mobile and you can use the same account on both so click that link in the description and you'll get 13,000 gold in a month of Premium subscription for free this offer only lasts 30 days so don't wait okay so yeah here's the plan I needed to see whether this AI stuff was actually progressing at an exponential rate is it getting that much better that much faster at the same time I needed to find out whether the general public would be against this whole AI art thing could I count on the bleeding hearts to stick up for me the artist and I was hoping that maybe there could just be something at I don't know intrinsically disgusting and inhuman about all of this all of this fake robot stuff and that people would just reject it they would reject the AI art and by extension The Starving Children and most importantly I needed to see how it would speed up my production process for my videos my idea was I was just going to start throwing in more AI pictures into my videos like with every new video video I'd have more and more AI art and see how people react and then I'd put all my findings into one big old video at the end of the year or now this is that video and I'm going to tell you what I found out number one most people don't care about this most people do not care about this stuff at all there are two sides to the AI debate there is the pro AI team you know the team that loves the AI the team that they love the robots and there's the anti AI team that they are the people who do not love the robots the pro AI team believes that this is a new technology that is going to lead us into the singularity where the robots will take over the world they're going to take all the jobs and then that would be bad but according to these people Michelle Obama will introduce Universal basic income so none of us will ever have to work again and then on the other side you got the the anti- aai team now they believe that this new technology will also lead us into the singularity and that the robots will also take over the world and our jobs but the the the big difference here is they don't think that Michelle Obama will introduce Universal basic income I tried a few experiments where I asked people to pick which art was AI generated and which art was human made and they couldn't tell the difference almost nobody could tell the difference just to mess with people I tell them that the AI generated art was made by humans and then they would proceed to tell me all the reasons why this this amazing Masterpiece was clearly crafted by human hands with emotions and feelings and then I'd show them something that was made by a real person and I'd tell them that a robot made it and then they would write out a paragraph describing to me all the reasons how they could clearly tell why this was generated by AI the next thing I found out is that the technology was actually getting better early AI art looked like this and by the time I made my first video on the subject it was still pretty meh you know like Spider-Man didn't have the right number of fingers Sonic was but over the the next few videos the next few weeks the next few months the technology was getting noticeably better Kermit the Frog close enough soon after this image to video generation started to get more popular you take a picture you shove it in the AI and a movie pops out for example remember that picture that I just showed you of Kermit the Frog imagine if it was moving wow that is that's moving number three none of this mattered out of the amount of time it takes to make a video like 70 for 75% of that time that total time is just me editing the video me getting the visuals me drawing the pictures me uh me zooming in the camera my fear was that if somebody could produce one of these videos with AI generated assets really quickly I'd be in a lot of trouble but it turns out AI is not as fast as I thought it would be yeah you can make anything you want you can get any picture you want but you don't really need to in most cases in almost any case actually for example if I want to use Mickey Mouse in a video under fair use law I have quite a few options I could go find an old animated clip of this stupid rat throw them in get like 30 seconds of usable footage from it if I'm feeling extremely lazy I could just find a PNG slap it in Jiggle It Around Ken Burns effect comedy gold but ideally I would just draw a picture of Mickey Mouse myself now this takes a while I'm a bit of a perfectionist so each Mickey Mouse drawing could take maybe 40 50 seconds so my goal in all of this was to see how much time the AI could cut down from these pre-established techniques how much how much time could I save by using the machines turns out none just generating an image of Mickey is going to take like a minute or two bare minimum and that's if they get him right so you might have to regenerate it a few times to get the perfect Mickey but even then that's just one Mickey with my drawings I can take I can take this Mickey cuz I drew him I could change his face you know change the background reuse this stupid little [ __ ] as many times as I want AI images don't work that way AI images are kind of a oneandone sort of thing you know and as time went on I started realizing that just using these pictures is kind of boring people like movements they like things moving around in these videos so I had to take these pictures and I had to get them to move the total process of all of this to get the AI picture the AI Mickey to get them moving to have something halfway decent to show it took a it took a while for a 4C clip Maybe 15 minutes this means to edit a 20-minute video entirely out of AI generated Clips would take about 75 hours after all of this I was ready to hang up my hats on this AI thing and so I forgot about all of this and just made a few videos about living under water but all that changed when I saw Sora is the video generation AI tool by open AI now like I said I have actually used you know a lot of video generation last here so this isn't exactly brand new stuff you know type in some words use an image get a moving picture this we we've seen it all before but the differences whereas the stuff I did was terrible this looks good it's not perfect but it is like a decade beyond what I was expecting to see this year that was concerning that was very concerning but turns out that's not all one of the other features of this Tech is video to video generation you can take any video that you've made and you can alter it using sora's tools this was something I also tried last year and it looked like it looked like this I look like a jackass sora's Tools looks like [Music] this one of the big things I've always talked about why AI is not going to replace art anytime soon is an issue of control when I make something in blender I have access to the X the Y the Z coordinates the key frames the materials I can perfect the timing an AI simply doesn't give you these tools so it's not all that useful in a lot of scenarios especially Beyond YouTube where you need that level of perfection for a Hollywood movie for a TV show but looking at Sora I start to realize it does have control the video to video generation is showing that the AI can figure out what the road is what the car is what timing is it has control it just doesn't give the user access to any of these tools it's showing an understanding of physics in cause and effect things that would normally require a game engine but it's just doing it with visual data it seems to suggest that the poten of this technology is a little little bit more than a lot of us gave it credit or at least I gave it credit I'm not saying it's going to replace anybody's jobs right now you shouldn't Panic don't go out there into the streets don't smash your roombas not stealing your job just yet but it will my fear of AI art has always been that at some point it will simply be the only way to produce economically viable art and I think we're getting pretty close to that if Sora actually allows for control of its Works which luckily it doesn't seem to right now but it probably could that would mean this sort of AI could be implemented in a traditional video editing or 3D modeling tool set let me give you an example okay let's imagine some advertising agency is tasked with creating a TV commercial to sell some tea bags it's called duct tea now the clever marketing team that has to come up with this ad realizes that they could feature a duck making and drinking some tea in the advertisement see the humor here is derived from the fact that ducks do not normally drink tea and the marketing team tasks one of their guys to draw up a quick storyboard based on this concept and here's what he comes up with door opens duck breaks into a house turns on a faucet fills up a glass of water turns the stove top on puts a teac kettle on the stove puts the water in the tea kettle there's a big explosion the teac kettle starts whistling woo woo the duck picks up the kettle uh he pours the the water from the kettle into the cup the duck takes the duck tea branded tea bags and puts them in the cup Stoops them for 5 minutes we see this quickly as the passage of time on the clock nearby starts moving you know it starts moving around and then the clock hits midnight and the duck quacks at midnight then he grabs the teacup he looks at the camera and he says go duck yourself see the humor here is derived from the fact that the word duck rhymes with [ __ ] normally to get this ad made we need a team of qualified artists each specializing in their own tasks it is not there's a reason why most local ads on the TV do not have do not look like this because it's it's expensive what if you could just take the story board that I just made take the the description that I I told you about several seconds ago you know take the audio of me saying go duck yourself and this concept art that I I drew you take all of that you shove it into the AI and it just comes up with the perfect 3D model based on the sketch comes up with the skeletal mesh comes up with the animations it infers details of the house based on your terrible drawings it it manages the camera angles creates the light sources gives you access to all the key framing data and positions of each object within the scene and with just a few tweaks you'd have a finished product you know you'd have the the ad would be done in like an hour or two something that would have taken weeks of planning and work something that would have taken a full team a long time to finish would take one guy one afternoon not even that this is this is not good I think we're pretty close to this stuff we're at least closer than I thought we were going to be and to be honest I'm not expecting any poison pills or copyright law to save the day you can't currently copyright any AI generated work but what I just described like a duck teabag ad that might start to count as significant human input or copyright might be something they can they might be able to copyright that and even if they couldn't who cares it's just an ad Forte it would save so much money and time but cost a lot of jobs I don't think people realize how much it cost to make animations and like commission drawings if I commissioned an actual artist to make this picture it would probably cost me more than this video you're watching will make an ad Revenue now imagine applying this mindset all these cost-saving measures across the entire competitive industry of art an industry we're Outsourcing to other countries is already common place to save a few bucks it's going to be rough look I had some hopes there for a little bit that maybe this we could avoid this AI fate for like another few decades we can't so uh with that with that said please don't take my job Indian Child once again huge thanks to conflict of Nations for sponsoring this video the free online PVP strategy game allows you to play the way you want to while trying to take control of the world during ww3 so click that link in the description and get 13,000 gold and a month of Premium subscription for free and remember this offer only lasts for 30 days so do it [Music] now [Music] he

Channel: KnowledgeHusk

Views: 100,903

Keywords: Sora, sora ai, ai to video, text to video, ai video generation, ai art, future of technology, tech news, openAI, chatGPT, future technology, artificial intelligence, AGI, video to video, trending tech, trending tech news, video essay, tech news right now, videos about technology, science, comedy, whimsu, knowledgehusk, knowledgehub

Id: fufr9VVVAcQ

Length: 16min 6sec (966 seconds)

Published: Thu Feb 29 2024