Script Pipeline

MI AI generated. This might be the most important video I ever make. I'm not exaggerating, but I'm actually kind of excited to make it. It'll probably always be a meme that I could be AI generated because I make videos about AI videos. And right now a lot of my videos are about how to detect video AI. So the most, you know, witty original comment is like, oh, about he's AI generated. And it's funny because I've done some testing and know AI video generator can actually quite do this. Some can do vertical videos, but they have very short record limits. The ones that can do longer videos are usually like avatar models, which I kind of refuse to try, which I'll talk about in just a second. And they don't look good anyway. I think that they look bad. So if you can do an AI generated version of this, I would love to know how. I mean, maybe you can, but if you send it to me, I bet it looks really bad. I'm not that's maybe that is a challenge actually. Maybe that's a challenge. Eventually, they'll probably be able to do it. So that's why I'm doing this right now because right now they can't in one day you might look back and be like, oh, yeah, that one time before models could do that, Jeremy said he wouldn't do it. So that's kind of cool. That's a cool opportunity that we have right now. I don't have any access to any specific AI models that the public doesn't have access to. So anything that I do, you can replicate yourself. I launched this channel six days ago. I know no one in the AI field. I know what I know about AI from using it and from reading white papers. That's about it. I think that you should try to use it if you feel like I could be AI generated. I really feel like you should try for yourself and you'll discover it's not quite that easy. I'm not going to edit this video. It's going to be an even longer, it's going to be a really long video. It's going to kill my algorithmic momentum. Holy cow, it's going to be bad. But it's important for me not to edit it. I don't want you to think that it's doctored, no audio edits, no trims, no graphics, no subtitles, nothing. I will never do an AI avatar of myself. I don't think you should either, unless you're really sure you want to. I believe that once you create an AI version of yourself that's photo realistic, people are going to be less inclined to trust you. I think that trust is very important. We're entering a no trust environment. Once these models continue to get better, people are just not going to trust anything that they see anymore. So how do you combat that? Well, you do things like this. You talk about how you're going to use it. I actually put an AI usage statement on my YouTube channel on the about. In eventually, when I have a website, I'll put it there too. Like I said, this is brand new. It actually states what I do and don't do with AI. For example, I do not write scripts with AI. I think that LLMs are really bad at script writing right now. Actually, I enjoy writing scripts. I don't use it to generate any likeness of me. When I do generate video models, I will always specifically state that I'm using them. I'll label them very clearly in the video and I'll usually put them in the credits if it's a longer form video. It's just important for you to trust me. And I wouldn't do something that would betray that trust because long term, we're going to need our trust. So if you see something and you feel like I'm not being genuine, I want you to call me out for it. It's important for you to state your own AI usage. In my experience, just in a few short days here, once you show that you know how to use AI, people don't trust you. Be careful. Don't betray people's trust. You're not going to always do something for them to not trust you. Just using AI, a lot of people don't like AI right now. That's totally fair. It uses stolen data. I don't think that the companies are bringing particularly responsible. The only reason that I take this seriously and I show how people how to use it is because I want people to use it responsibly. People are going to use it. That's just been my philosophy on a lot of things in life. I'd rather people do it smartly because the tools are out there. People are going to use them. And my AI usage might change and I will make sure to specifically state that if it does. But I'm never going to generate an AI avatar of me because I want people to trust me. And I think it's creepy too. When I uploaded a photo of myself in my Google Vio review, which is the first video, first long form video I posted on my YouTube channel, I thought it was creepy. I thought it was really creepy. I mean, first of all, it didn't quite get my likeness right. So it looked, it just looked bad. But then I asked myself, oh, why did I do that? Like, cool. I need to test it. But I don't think that was worth it. I don't think that was worth it at all. So I think this is long enough. I think it did its purpose. Here's what I'd like to leave you with. As we're entering this period of no trust and people don't believe what they see anymore, I encourage you to do something similar to what I'm doing right now. Whether that's an AI usage statement in some place that people could see it, whether that's just telling people what you will and won't do. So, you know, AI can probably do scams really well right now, right? So having like, safe words with family to tell them, yeah, if I ever actually do need help, this is how I'll ask it. You know, putting things in your website would be a go-to, just figure out a way that works for you. Because we're entering what I think might be a pretty bad period. And in order to avoid it from being such a bad period, we need to hold AI companies accountable. Google has this tool called SynthID, which is supposed to detect their own AI videos with like invisible fingerprinting. Of course, they release that after they release the Vio, because Vio makes them money, whereas AI fingerprinting doesn't directly make them money. So like the profit motive is too strong. This is going to be a very, very common problem. We need to hold AI companies accountable for it, just like we need to make sure that YouTube and TikTok and other platforms like this that could have AI generated content have strong guidelines on when they can and can't use it. Right now, both YouTube and TikTok require you to check a box if something's AI generated, just to like encourage disclosure. I don't know how they're policing it, or if they are, I haven't tested that, and I'm not about to test that. But eventually, hopefully we figure this out before they get better. I do think that you can still spot most AI videos right now. I do think that there are giveaways. But once, if and when the models get better, we're going to need real tools that aren't just looking at AI videos or me making a video being like, I'm not going to do it. We need real solutions, and I'm just not smart enough to do that. I'm not a developer. I'm not an AI researcher. So I'm just here to help you figure out how to use it. Thank you for watching. Keep it real.