AI-generated art is impressive, this picture above is generated by DALL·E 2 and shows the quality that a user can get if the AI is appropriately trained on a large enough selection of images to get an idea of what a certain object looks like. It seems to be getting better and better as time goes on. AI can be used to generate videos with a talking avatar, using a text prompt. This is indispensable for placing instructional videos on a website, and the process of making the videos could be automated. The Synthesia website can do this easily, no longer will you need a model and the cost and time involved in making videos for instructional or demonstration purposes. Now it is much easier to accomplish this with a simple website. Now anyone can become a video creator and add great content to a website with less fuss than before. I do wonder how you could make yourself an avatar in an AI-generated video, but it can surely not be too hard to do. Instead of using placeholder avatars, you can make your own.
Synthesia allows a user to do this, would it be good enough to make instructional Youtube videos without an expensive camera and studio setup? I guess you could. A good feature would be removing the background to allow chroma keying and making something else the background. I think a lot of the videos on Youtube might already be AI-generated, easily churned out for clicks and likes. You can buy subscribers and commenters on Youtube already, so this would not be surprising. But this is not a good idea in the long run. Better to just create good content and then the subscribers will come.