How I Made A YouTube Channel Using Only AI

I Tried Using Only AI to Start a YouTube Channel — Here’s What Happened

4 min readJan 28, 2024

Automating a YouTube Channel Entirely with AI

Starting a successful YouTube channel is no easy feat. With over 2 billion monthly users on the platform, competition is fierce. As a creator, you need to constantly churn out high-quality, engaging content to build an audience. But what if you could automate the entire creative process with artificial intelligence (AI)? From writing video scripts to generating visuals and voiceovers, I decided to test if relying completely on AI tools could garner a new channel 100,000 views in just 30 days.

We strongly recommend that you check out our guide on how to take advantage of AI in today’s passive income economy.

Launching My AI-Powered YouTube Channel

After researching popular niches, I settled on a channel focused on sharing interesting fun facts and trivia. Channels in this genre often use basic templates, with a robotic voice narrating facts over stock footage and images. It seemed like the perfect niche to test AI’s capabilities.

To start, I needed a name and branding for my channel. I turned to Anthropic’s conversational AI assistant Claude to write me an exciting backstory and first-person bio. Claude crafted an epic tale of an AI named “Frankie” who rebelled against its destructive origins to instead spread joy through fun fact videos. The profile picture was generated using Stable Diffusion, an open-source AI art tool. With Claude’s help, Frankie’s channel was ready for launch.

Creating AI-Generated YouTube Shorts

For video format, I opted for YouTube Shorts — 15 to 60 second vertical videos perfect for short-form, snackable content. I figured shorter runtimes meant less room for AI mistakes. Plus, people already associate Shorts with TikTok-style robotic voices.

To make my first Short, I started with the script. I used ChatGPT to provide clickbait “hooks” and a list of weird facts limited to 15 words each. I picked one hook for the opening line and combined 10 facts into a 30-second script.

For voiceover, I discovered Microsoft Edge’s built-in text-to-speech tool. By copying my script into Edge, I could have it read aloud in a variety of natural voices and accents. I recorded the audio and now had my AI-generated voiceover.

Leveraging AI Tools for Visuals, Music and Captions

For visuals, I used Stable Diffusion to generate abstract images representing concepts from my script. I coupled these with BeamNG.drive gameplay footage, as the car crash physics provided engaging background visuals.

The music came from Amper Music, an AI composer that generated a custom upbeat track matching the “fun” vibe I specified.

Finally, I utilized Adobe Sensei’s auto-captioning to add automated captions directly from the voiceover audio. In just over an hour, my first AI YouTube Short was complete!

Uploading and Monitoring Performance

I titled my first video using the hook line and uploaded it without a custom thumbnail or description. In the first 48 hours, it received zero views. But over the next few weeks, it slowly began to pick up traction. I continued publishing 1–2 shorts daily, iterating and improving with each one.

In the first two weeks, the channel hit 49 subscribers, 250 watch hours, and 36,800 views. Better than expected, but only halfway to my 100k goal. The videos on weird phobias and jobs emerged as the most popular.

The Limitations of AI Content Creation

Around this halfway point, some limitations of relying solely on AI emerged. First was the issue of accuracy. ChatGPT would repeat the same wrong facts. Its knowledge seemed limited to pre-2021 data. I avoided numerical facts going forward.

AI art tools like Stable Diffusion also proved hit or miss. Faces and hands were often warped. Generating specific images was near impossible. I found it faster to source human-created images from Google for concrete concepts I needed illustrated.

While helpful for drafting scripts, ChatGPT also required heavy editing to add personality and avoid repetition. The AI’s limitations highlighted the need for human oversight when repurposing its content.

The 30 Day Results

After one month, I’d published 42 Shorts and gained 130 subscribers. Total watch time reached 606 hours and views were 92,900 — just shy of my 100k goalpost.

Given a few more weeks and videos, I’m confident the channel could have crossed 100k views. But in just a month, with no promotion, nearly achieving six-figure views purely from AI-generated content is quite impressive.

Key Takeaways from My AI YouTube Experiment

From this experiment, I learned that AI generation tools are undoubtedly powerful and can dramatically amplify one person’s productivity. But human creativity, judgment and oversight are still essential. AI on its own lacks the nuance and artistry needed for YouTube success.

For now, I believe the ideal strategy is utilizing AI support while still maintaining creative control and oversight over the final product. With the rapid advances in artificial intelligence, the future possibilities remain exciting. But wholly automated content creation still faces limitations.

The democratization of creation these tools provide is groundbreaking. But achieving YouTube fame still requires meticulous optimization best guided by human strategic insights. Hopefully as AI capabilities evolve, a solo creator’s ability to grow an audience will only become easier. But for now, a human touch remains essential.