AI video tools: not quite ready for primetime

Not sure if this is 100% on topic for this group, but thought some people might be interested to hear my experience.

AI video tools have been getting a lot of hype in wake of the chatGPT frenzy. I was curious if I could leverage a series of AI tools to make a no/low effort YouTube channel.

The idea

1. Find a YouTube niche using vidIQ
2. Use Looka for logo inspiration
3. Set up chatGPT to spit out a video script given a topic/input.
4. Feed the output into a one of these tools
5. Feed the video into the YouTube API to be released on a schedule
6. Queue up a bunch, sit back and see if it gains any traction

How it went

1. Though it looks very good it turns out vidIQ needs you to have an existing channel to work from so I ended up using traditional tools like SpyFu. I found Animal Fun Facts as a low competition area that I knew chatGPT would be good at, and a low production value wouldn't hurt too much.
2. The logo was a breeze. I have a couple nitpicks about not being able to select fonts by name and one or
3. Setting up chatGPT took a bit of prompt engineering to get a high quality result (remove queues/labels, format consistently, more interesting facts, etc.) but the final results were very well laid out and the facts were really interesting for weirder animals.
4. Tools I tried were Kapwing (very accessible demo) and Pictory (best reviewed tool).
   • Kapwing just takes a phrase and makes a video. This means I couldn't prompt engineer effectively and script quality was much lower. Backgrounds were images instead of video, and backgrounds occasionally did not match subject material. This would not cut it.
   • Pictory was very disappointing given that reviews are generally positive. In my case I found that it only got the content right for about 40-50% of "scenes" (auto generated, roughly 1 per sentence of script). In the end I found this unusable even when trying to select correct content manually because the stock footage library often did not have 20+ items for the animal I wanted, and I didn't see a way to use the same content for multiple scenes.
5. Stopped there because I couldn't see a way to get usable video, but I believe the rest would be relatively easy.

tl;dr
Tried to automate a YouTube channel with AI tools. Tools for research, logo, and script were all quite usable, turning out pretty good results. Video creation itself needs a major leap before it is usable without adjustment. Needs improvement to even be useful as a starting point for this use case.