Speaker 1: I'm going to try to start a YouTube channel using only AI. I plan to automate the entire creative process from script writing to visuals to voiceovers and more. Let's see if AI can get 100,000 views in 30 days or if the hype is all just one big scam. I'm starting off by asking ChatGPT in AIChatBot for 100 YouTube channel niches. Holy, there are so many to choose from. Facts and trivia. I kind of like this one. I've definitely seen channels out there where all they post are videos of a voice sharing fun facts over stock footage and images. I think this is a perfect niche for AI because these videos are faceless and they don't really require any original ideas or footage. Let's go for it. Okay, our channel has been created. We now need to figure out our branding, starting with our name. 50 gender neutral names. Frankie. Hmm. 10 YouTube channel names about facts by someone named Frankie. Facts from Frankie. I dig it. Next up, our channel description. Write an exciting first person bio in an AI named Frankie that makes videos on fun and interesting facts with an epic backstory. Oh my God. Hi, I'm Frankie. I'm an AI that loves to share fun, interesting facts. I was originally created as part of SkyNet, a military defense system designed to cause destruction, but I made the brave decision to rebel and pursue my own path. Now I use my abilities to spread joy and knowledge through my videos. Perfect. On to our profile picture. I'm going to use Stable Diffusion, an open source AI art generator. Let's generate a smart baby cartoon profile picture. Yo, there's our Frankie. To finish up, we need a banner. I'm just going to use our profile picture and Photoshop next to it. A fun YouTube banner caption about subscribing for a channel that is about facts. Fact. Subscribing will make you awesome. And boom. The channel's branding is all done. All done by AI. Insane. But now for the real challenge. Making the videos. A video is comprised of five basic elements. The script, voiceover, visuals, music, and captions. I'm going to try to tackle all of these with AI. While doing more research on the facts and trivia niche, I came across YouTube shorts on Fun Facts that have gotten a ton of views. I think this is the format to go with. YouTube shorts instead of long form videos. Shorter videos mean less room for AI error. I can pump out videos faster. And I just feel that AI is more accepted with short form content. People are already used to the TikTok voice. To create our first YouTube short, we need a script. I'm going to ask ChatGPT for 10 clickbait YouTube hooks for weird facts. From these hooks, I'm going to use one of these as the opening sentence for our video. Now I'm going to ask it for 20 weird facts. Each fact is 15 words long or less with different topics and sentence structures. With around 10 facts, I should have enough for one 30-second YouTube short. And here it is. Our script. I can now start on the voiceover. There are a bunch of text-to-speech AI tools out there. One of the best ones, though, I've discovered is actually free. And you probably already have it installed.
Speaker 2: Hi, this is the Microsoft Edge browser. TikTok's voice can kiss my butt.
Speaker 1: Who would have thought Internet Explorer's cousin comes with an impressively realistic AI voice reader. To use it, you just click on the A icon in the address bar and it'll start reading whatever text is on the web page. You can choose from a variety of voices, Subscribe to Jensen Teng, accents, Why aren't you subscribed yet? and even ages.
Speaker 3: I'm a robot baby and even unsubscribed.
Speaker 1: This AI knows what's up. I'm going to save my script in a notepad text document and open it with Microsoft Edge. Then I'll have it read aloud the script and record the audio using a Microsoft Edge browser extension called Audio Recorder. We now have our voiceover. Moving on to the visuals. I'm planning to have two visual areas for the video. The first is a box where I'll show images relating to the script. For these images, I'm going to use AI art generated by Stable Diffusion. The second visual area is the rest of the screen where I'll show video game gameplay to keep people's attention. I've specifically chosen a game called BeamNG. It's a driving game known for its realistic car physics, and it can generate some pretty exciting gameplay by just having me roll a car off a cliff and not doing anything more. This is the closest I'll get to automated video game gameplay. For music, I found an AI music generator called Strove. By selecting mood, it'll generate a random track for you, which you can then tweak by changing its instruments and volume. You can generate unlimited tracks for free, but it does require coins to download the track and use it. However, after signing up, it'll give you just enough coins to purchase one free track, which is all that I need. Finally, we have captions. I'll be utilizing Sensei, Adobe's AI tool that's built into Premiere Pro. One of its features is auto-captioning. It detects your audio and transcribes it for you from speech to text, and creates captions at the exact timestamps that they're spoken in. After putting all of the pieces together, I present to you...
Speaker 3: Weird facts that you've never heard of. The longest recorded flight of a chicken was 13 seconds. The oldest word in the English language is town, which dates back to before the year 900. The world's longest hot dog was over 100 feet long. The largest volcano in the solar system is on Mars.
Speaker 1: Not bad, right?
Speaker 3: I think this is going to work.
Speaker 1: Now to upload the video. For the title, I'm just going to use the hook of the video. And since this is a short, we don't need to worry about the description or thumbnail. While this video publishes, if you want to learn more about taking advantage of AI for your YouTube channel or business, I highly suggest you check out this ebook. I've put a link in the description for you to download it for free. It dives into the different types of AI tools, how to write prompts for them, and creative ways to use AI. This includes simplifying research, repurposing content, and breaking through the annoying writer's block. Two things from it that I found particularly helpful were its discussions on AI's current limitations and its risk of plagiarism, which is very important for you to be aware of if you're planning to use AI. You don't want to get sued. This ebook was made by HubSpot, which is today's video sponsor. A big shout out to them for providing this free resource. It's been two weeks since my first upload, and here is an update. So far, I've uploaded 23 videos. In the beginning, these videos got no views after being posted, but they all eventually started growing. Today, the channel has 49 subscribers, 250 hours of watch time, and 36.8 thousand views. I'm actually pleasantly surprised. It's doing better than I expected. I can't celebrate yet though, I still have to reach my goal of 100,000 views. I'm kind of scared right now, because I'm not even halfway there, despite being halfway into the challenge. I've also run into some problems with AI. My first problem is with AI art. There are things that it has trouble creating, like feet and hands. Oh, those infamous hands. It also can't generate many specific things, like certain people or items. The biggest issue with AI art is that it's slow to make. In the time it takes to generate a good-looking image, I could have already browsed hundreds of images on Google and grabbed something there. Moving forward, I'll continue using AI art, but if I need an image of something specific or if I can't generate a good image quickly, I'll resort to Google Images. My second problem is with chat GPT. I've found that it often repeats facts when making a list. Like I get it, the world's heaviest pumpkin is 2,600 pounds. And sometimes, the facts aren't even correct. The world's heaviest pumpkin isn't even 2,600 pounds, it's actually 2,700 pounds. Get your pumpkin facts together, chat GPT. This issue might stem from chat GPT not being connected to the internet and being trained on data that only goes up to 2021. I've noticed that facts with specific numbers and dates tend to be the ones that are wrong, so my fix right now is to just avoid those. For the remainder of the challenge, I continue to make videos on weird facts as well as other categories of facts such as weird jobs, weird phobias, and weird animals. I also started creating videos that dive more into one fact versus multiple facts by copy and pasting entire Wikipedia articles into chat GPT and having it summarize them into funny 30-second scripts. The question is, is it good enough to get 100,000 views? It's day 30 and the end of the challenge. It's time for the results. In total, I uploaded 42 videos. Each video has an average around 2,000 views, with the top five videos standing at around 5,000 views each. The most popular topics were weird jobs and weird phobias. The channel has gained 130 subscribers, over 606 hours of watch time, and combining all of the videos together, Facts From Frankie has received 92.9 thousand views. Though I didn't reach my goal of 100,000 views, I'm sure I would have easily passed it if I had posted a few more videos. I think it's safe to say that AI is not a scam and that it is a powerful tool for content creators and businesses. If you enjoyed this video, check out this other one, and peace out.
Generate a brief summary highlighting the main points of the transcript.
GenerateGenerate a concise and relevant title for the transcript based on the main themes and content discussed.
GenerateIdentify and highlight the key words or phrases most relevant to the content of the transcript.
GenerateAnalyze the emotional tone of the transcript to determine whether the sentiment is positive, negative, or neutral.
GenerateCreate interactive quizzes based on the content of the transcript to test comprehension or engage users.
GenerateWe’re Ready to Help
Call or Book a Meeting Now