Veo 3’s Visual Uprising: AI’s Video Magic and the Shift Back to Text

E70 VEO 3’S VISUAL UPRISING_ AI’S VIDEO MAGIC AND THE SHIFT BACK TO TEXT

Transcript

Dalton Anderson (00:01.358) Welcome to VentureStep podcast, where we discuss entrepreneurship, trends, and the occasional book review. Today, we will be discussing the recent release of VO3, and this is a model created by Google and has been a really good model and industry leading, but some recent releases make this model very captivating and hyper realistic and potentially unsettling, depending on who you ask.

But I think the general sense of the VO2 release is wow, AI is advanced and AI is moving much faster than I ever anticipated. And I thought this would be something that I wouldn't have to worry about. And it wouldn't be within my lifetime that we would see something that would potentially affect me because people have been talking about this AI, AI doomer gloomer.

type of vibe for a while, but then people kind of put that on the back shelf and they're like, it's whatever. Like people talk about AI all the time. Like I've heard about it for years and it's not doing anything. Look at it. It still sucks. And then they see this VO three and they're like, my goodness, what is that? That's an abomination. That's insane. What is that? That's not real.

And I'd showed this to a friend and gave him very little background about it. Showed him this video. was like, hey, what do you think? And then later on, I told him that it was AI. And he's like, wait, wait a minute. Where are we getting, where are these people from? Where are the people from? He kept asking. And I was like, what do you mean, where are the people from? He's like, well, I don't know. Where are the people from? Are they getting them?

They're getting them from other videos. They're taking them and then putting them in the video. Are they getting these people from YouTube or where are they getting them from? And I was like, they're not getting them from anywhere. And he's like, no, they're getting them from somewhere. I was like, no, these are all AI. Like this is all AI. I mean, it's trained on videos that has people in it, but these are all unique people for this video. And he did, he was like, nah, man.

Dalton Anderson (02:29.9) No. And I was like, I don't follow. And he was just like, this doesn't make any sense. This doesn't make any sense. And yeah, it was just a funny interaction. So I showed him another video and he was like, bruh, that's insane. And so I think people are starting to realize like,

how advanced this whole AI thing is. And I know that if you're listening to this show, you know, and if you're new here, then you'll know over time, but AI is much more advanced than people are giving it credit. I mean, it still has limitations for sure, but it's progressing at a rate that is unprecedented for many technologies. And it is quite impressive. And so in this episode, we're going to talk about like a little intro of what Vio3 is.

I wanna show some videos and what I think is my hypothesis and thesis is that since that AI is progressing at a rate so rapidly, aren't gonna be able to know and people already have a hard time differentiating what is real and what is not real. But then when you add in these AI components and this ability to iterate fast and tell stories, then it's a slippery slope for scammers and people that are

miscontruding the truth, people are going to become, in my mind, and this is my thesis, people are going to become more centered around text and ID verification. And I'd mentioned this before on my episode 63. It was the, oh, episode 63, I had look, The Imperfect Echo, AI Cloning and Its Current Limits.

And in that episode, I had cloned my voice and for the first seven or so minutes of that episode, that wasn't me. It was a version of me, but it was artificially generated. And unless I really told you that it was AI generated, people didn't know. And I knew because I didn't do it, but there are...

Dalton Anderson (04:52.686) Once you find out that it's AI generated, you could pick up on little things in that video or not video, but the audio that will somewhat give away that, this was AI. But those are the things that we'll be discussing this episode. And of course, my name is Dalton Anderson. I'm your host today. I am doing all sorts of things in life from data science, programming, insurance, and building companies. So.

Not building my own yet, but I'm building things. So I'm a builder and an explorer. I like to read, run, and do lots of other things, but also podcasts. So here I am. Okay. So we'll get into it. So VO3, what's the difference from VO3 and VO2? So VO2 was incredible. It was really good, right? And it was the best. It was the best on the market. But the fundamental issue with VO2 is that

The video durations were short. The ability to have native sound and voicing and consistent characters. So those are a couple crucial items. the voice piece, there wasn't a robust solution for once you made, for the after product of a VO2 video, if it involved voicing, like voice acting.

So if you were to do that, then you'd have to figure out a way to get the video to speak but without sound and then voiceover something or use AI. These are AI products that will make a voice for that person. And then the other piece that was an issue was consistent characters. So that's a bit odd, but consistent characters are important when you're telling a story, right?

person that's doing something and you want to tell a story like they go to work and then they come home and their house is burned down and they're devastated. And there's photos of them driving from and to work. And then there's photos of them sad and maybe they build a new house from scratch or something like that. It takes them a while and it shows the hard work and determination that only works if it's the same person every time.

Dalton Anderson (07:19.52) And if there's not the same person, then it's a little bit more difficult for people to be captivated by the story because people attach themselves to certain characters in the story because they find them interesting and want to see what happens next. But if it's a different person on each scene, they're like, I don't really care. So you needed a consistent person. You needed voice acting or voiceovers or something like that. Voice generation, really, for the video.

before you needed voice overlays, now you don't. And then now also it generates the sound effects in these different things. And so it's pretty insane actually, like legitimately nuts. And I have some really cool videos that were put together on X, the X platform.

And if you are listening into the show and you're on Spotify or YouTube, you can toggle to the video option and you can see what I'm sharing. But if you are not on those platforms, you're going to be listening in. I'll make sure to share the sound and also I'll make sure to copy the video links and put them in the show description. So if you're like, wow, that that really sounds interesting, I want to watch the video myself.

I'll make sure that I copy and paste the video links in the show description. Okay, without further ado, let's start sharing. So the first one that I'm gonna share was a video that my friend Tom sent. Shout out to Tom. I don't know if he wants me to plug his last name in there on the internet, but shout out to Tom. He's been on the show before. So I guess you would know his last name by Tom Mercer.

He provided this video to me and was like, have you seen this yet? And I was like, no, I haven't. And this is good. So I want to share it with everyone else.

Dalton Anderson (09:28.608) All right, so let's do this. So this video's title is This is Plastic, made with VO3. All right, and it's by Metapuppet, which he's made some pretty good videos with his other friend, but it makes sense that he's part of this AI studio that is using AI to build original content and or using

live actors, but they also have made just solely AI content series. And this is one of them, which blew up kind of, it's got almost a million views just on X and I know it's been shared at multiple platforms to have a lot of views, but without further ado, let's get the video started.

Dalton Anderson (13:04.408) So good.

So that was an original AI generated content only made with VO and some clip. I assume that they made like a prompt for like one character and like of a closeup of that character. And then they had to do several things within that clip. And then they cut that clip to make a.

multiple shots with like they would get multiple shots with one character and then they would clip that video that's generated multiple times to plug and play into the content that that they created but that was really well done by MetaPuppet and then there's also this other one by Hashim Al Ghali I think I pronounced that

And this one is the afterlife, the unseen lives of the AI actors between prompts made with VO3. Before I do that, if you didn't follow what that video was about, the last one, and you were just listening in, it's a little bit difficult to tell about the beginning part, but I think you pick it up later on in the video. But if you didn't, it was basically, there was something on TV that says like, know, microplastics are

found in men's testicles and a man's drinking out of a water bottle. And then he's like, you can't believe everything you hear. And then his partner is pregnant and their baby is plastic. Like it's a plastic baby. And that's when you were hearing, yeah, he can't play baseball. Like if he gets hit, he'll get all dented. He'll break. And everyone's laughing at him and all these, all these references to where we have

Dalton Anderson (15:02.87) in society, well, keep plastic off the beach and like the baby's at the beach and like he's just sad. And that's why at the end of the video or not the end, but the end of the marketing video and then in the boardroom, they're like, I just felt bad for him. And that's why, because it was just, he was just getting dogged. So, but now let's share this screen.

So we want to share the afterlife.

Dalton Anderson (18:35.982) It's pretty well done. The music in the background reminded me of the Westworld intro and overall a captivating story about how, what happens, I guess what happens to the actors that are created when you create a scene and they go to the void and they're not too happy and

there's some actors that are repeating scenes where like they're killed and they don't want to do it anymore. Yeah, no, it makes sense, but it was pretty good. And then to bring it all in into a vacuum, there is a studio and it's by Dave Clark. He's the co-founder and the chief creative officer, Promise AI. And that metapuppet person that made

plastic video is also part of this Promise AI studio. And I think he's like the creative director. And then Dave Clark is the chief creative officer. But I wanted to also share, okay, it's not all doom and gloom, right? Like I'm not a doomer by any means. I think it's never been more exciting to build today than it has been ever.

ability for you to have a voice and to do something that you're interested in, you've never had this ability before. If you want to code something up, you can try it. It's probably not very good because if you don't know anything about coding, it's going be really hard and you won't know how to ask the right questions or the prompts. But you can take a class on that and you can learn how to code. You can look at stuff on the internet, read books. But I think it's never been easier to get started.

and remove that barrier of entry. And then once you've got something that kind of works, then you can learn the detailed craft of the code and the reviews and understanding the little things. But coding, like the big things, you can kind of use AI to do that. And same just here. If you're a creative person, you want to get into music and you don't have any idea how to do that.

Dalton Anderson (21:01.91) Well, AI can help craft some beats for you and you can just hop on a song or vice versa. You could be the producer and you can have an AI artist make lyrics and sing for you over the beat. Same thing with videos. This is a video that was created with

with, how do I phrase this? What I was saying before was, there's a good way to get started, but if you're an expert, you can use AI to help you. And so like I use AI to help me code. I'm not saying I'm a expert at coding, but you can use it as an assistant or someone that's an advisor on your project and can help you out. And that's an example. This is a good example. This NinjaPunk,

IP that's being created at Promise AI, which I'll be sharing my screen in one second, is a video that was helped. It was helped. that doesn't make sense. A video that was AI assisted. And what they said was that there's live actors and then there's also

AI built in with the live actors to enhance the scenes and they really focus on using AI to generate characters and world building and then help create costumes and these other things. But this is AI partnering with live actors and the results. Chef's Kiss. I really like sci-fi stuff, but it just seems that to build the worlds for sci-fi, it's

pretty complicated because the worlds are typically vast. There might be multiple planets and each planet's got their own little country or not country, but own little culture or maybe not their own race, but culture and or like environmental changes that happen to people's bodies. And it's just a little bit more complex than like a normal show like The Office. I I'm I'm not dog in the office, but

Dalton Anderson (23:19.63) I mean, you don't need much to make the office, but to build a captivating sci-fi show or movie, there's quite a bit of time spent on the world building. But if AI can help with that, can expedite, not expedite. my gosh, I'm messing up today. This is all a a word ramble today this morning.

it can speed up the process. So I'm going to share my screen here. And this is of Promise AI's original IP. And these are the type of people, by the way, that I want to get on the show. This is really interesting. And I would really like to know how they did it and how they plan on partnering with AI in the future and what other cool stuff are they working on. Because of the really good videos, these AI videos, it's coming out of this place, like

like the three or four, I didn't want to share a podcast and have like, you know, 12 AI videos, but Dave Clark and Metapuppet are the folks that were building really cool AI videos. There's some other people, of course, but they had a majority of the like crazy insane AI videos. So would like to get these people on the show.

they're willing. I'll ask them and see what they say. So let's do this.

We are.

Dalton Anderson (24:55.31) Okay.

is this the wrong one?

Dalton Anderson (25:02.712) Here we go. this tab is in. Sure in the wrong tab. All right, so this is this original IP that I was talking about.

Dalton Anderson (26:18.03) Okay. Hold on.

All right, so that was this AI partnered original content generation. And it's really good. Incredible. It's incredible. It scratched that edge of altered carbon. Yeah, very good. Love that show. I wish they made another season. Unfortunately, they just got rid of it, I think because it was too expensive for what it was doing for Netflix. And so there's some shows that

maybe have less of an audience and low budget, it's fine. High budget show, low audience, not so much. So it was canned, unfortunately. But really good show. If you haven't watched it, I would watch it.

So of those videos I showed, I think a lot of people are going to be concerned about, okay, well, how do I know what is real or not? And as I mentioned, I talked about that on episode 63, the Unperfect Echo AI voice cloning and its current limits.

because this is where we are right now, but where will we be in two years? Two years is not that long. And where we are now is where people think that we would have been probably in 15, 20 years. And like the timelines of which people are predicting these things to progress and the rate that they're progressing are way off, way off. And people

Dalton Anderson (28:00.448) weren't paying it any mind. And I think now it's a bit surprising how well these videos are being created. And people are like, wait, hold on here. I thought AI was something I didn't have to worry about until later. And that answer is just not true. And if people are saying that they're lying, like there's the wave is coming and you're either riding it or you're

you're showing away or something. the premise of the episode is basically to illustrate that VO2 is cracked. VO3 is cracked, VO3 is cracked and this is something where it's getting a broader audience. Whereas the little niche groups of the AI folks or the tech folks,

They understood it, right? But this is more of a general public video consumption and it's freaking people out. People are getting freaked out and rightly so. And I think long-term.

think long-term there's gonna be two fundamental shifts potentially. And this is my thoughts, not true. It's true that it's my thoughts, but I don't know if it's gonna be true in the future. So one, I think there's gonna be more of an emphasis on ID verified Instagram accounts. And I talked about this before and multiple times, either in the prior episode, the Imperfect Echo or the filter bubbles

in that episode about the AI models that they were making or AI models that were on social media. And like, how do you know who is real and who's not? Why I do these video podcasts? Because I felt that in the future, like three years from now, like, and I'm saying this like a year and a half ago.

Dalton Anderson (30:03.724) that the AI is gonna become so advanced that it wouldn't matter if a person was doing the podcast or not. And people are just gonna spin up these podcast farms and make podcasts. And then people are gonna be really uncertain on like who's a real podcaster and who's an AI podcaster because well, I don't think that it's worth my time to listen to somebody and listen to their thoughts if they're just this.

AI generated prompt. Like I can do that myself. I don't need to listen to an AI bot on, my podcast. So that's the main reason why I did videos because videos, you can verify that you're real, that it's me and I've been doing this for a while. And there's this whole bunch of things that like videos also a little bit more engaging that that's why I felt that it was really important for me.

to do video because long term I think AI is going to be so advanced that people aren't going to really trust broadcasts as much unless they're videoed or whatever it may be or verified. And I'm saying that because I talked about it quite a bit, but in a general sense, there's going to be a shift from these anonymous social media presences to a more professional presence back to when people are a little bit more serious about social media.

maybe like 2012 to 2016 where people had their first and last name as their Instagram handle or Twitter handle. That I think is gonna shift from where it is now, where it's shifted from professional presence to anonymous presence. And then I think it's gonna go back to professional presence because people aren't gonna trust some anonymous account. They're just gonna think it's a bot. And with so many...

so many bots and or real people that are determined to be bots. And if you don't want to be determined to be a bot, then you're going to have to make your name like your name. And then you're also probably going to have to ID like verify with your ID. I know X is doing that for their premium accounts and they have been for some time. And I think that was one of the main points that Elon was making was, hey, we've got so many bots on the pro on the platform and

Dalton Anderson (32:28.212) Majority of the activity on X are bots and so we've got to cut down on all the bot activity and I think at a certain point like They you know, they got rid of their API and they got rid of these other things and then they're like hey if you want premium then you have to verify with your ID People didn't like that, but I think it's the right direction long term I think Instagram is doing something similar for their premium accounts. Like if you wanted a premium, then you have to verify with your ID

So it's becoming increasingly more common. And then the next thing that I think is a general shift is I think the preferred medium of consumption is gonna shift from, and this is for social media, not for TV or these other things, is gonna shift from videos and images, act to text, I think for casual videos, like casual videos, casual images.

I'm not saying for YouTube or documentaries or these other things that are not part of the general TV of Netflix to HBO. think it's called HBO Max now. on, let's just spend a moment on that. I think it was HBO Max, HBO Max Plus, HBO Max, and then Max something, and then back to HBO.

Max, like what are they doing there? Anyways, but those videos that don't sit with either YouTube or with...

like these large media companies, and YouTube's pretty big as well, so it is a media company, but you know what I mean. If it's more of like a short form video, I don't know. I think people years from now are gonna have a hard time differentiating what is real, what is fake. And when everything you see is questionable,

Dalton Anderson (34:43.352) then what's the point? So I think, and this is what my theory is, is that people are gonna become more comfortable with text because text is fine and dandy. If AI generated text versus AI generated the video, an AI generated video is way different than AI generated text. AI generated video, it's like the whole thing is,

The whole thing is just...

nothingness. I don't know. it's not trying to be rude to AI video. mean, that one ninja punk thing is really cool. But I'm saying for like a casual video of like, I'm at so and so place or this and that. just don't understand like the engagement of these AI videos. If everything that you're seeing on social media is just like these AI videos that were created, then what's the point? You can also see it as

Maybe they make memes with it. I think AI memes are pretty good, but.

I think people are going to become more comfortable with text versus videos and images in a casual sense and looking at things casually. But maybe I'm wrong, but I think that people are just going to become very wary of like random things that they see on social media now because they just, they don't, they don't trust it. They don't trust anything. You don't know who your friend or foe is and

Dalton Anderson (36:24.002) You don't know if your buddy created this video or if your buddy said AI to create it, like, you know, with a prompt. So.

Dalton Anderson (36:39.232) I do think that maybe five years from now it might shift back where people come more comfortable with everything. But I think there's gonna be a knee-jerk reaction when these things start flooding social media and people are sending these things to each other like, my gosh, you saw what just happened? And then later they find out that it's an AI-created video. And then after that happens a couple times, they're like, all right, you know what?

I'm out. I'm not doing this video stuff anymore. I'm not doing these images. I can't trust this stuff. I don't trust it. And it might, it might shift back to like traditional media and it might give traditional media and or X a stronger foothold in people's media consumption rotation because you can trust that these things were created by

person, like really spoof-readed, but I know that these news companies are using AI to make content. So, yeah, it's gonna be interesting how it all plays out. I I could see it going two ways, where people just say, I'm out, I'm not interested, or they really get into the memes, and or the third option is that they move towards like traditional media and

Not necessarily like newspapers, but not the paper format, but still like digitally looking at this stuff. I think those are the three outcomes. Not sure which one, but I do think that there's gonna be a knee jerk reaction with all these AI videos starting to flood social media and the inability to tell what is real and what is fake is not good.

especially if that's not what you're there for. And it's not gonna go well with a lot of people. But I hope that you found this video exciting. Once again, I'm gonna put these, not videos, this podcast, I hope that you found this podcast exciting. I'm gonna put these videos in the show description as I mentioned before. And as always, wherever you are in this world, have a good day, a good morning, good afternoon.

Dalton Anderson (39:04.216) Good evening. Thank you for tuning in and I'll listen. Listen, I hope that you listen in next week. I've been all over the place today, but anyways, goodbye.