I momentarily though that the biggest reason video tutorials are going outcompete text for at least the next few years is that it's currently much more difficult for AI models to turn video into plausible semantic mulch than text, but then my brain said "oh, that's not a problem, you mulch the transcript text, use a text-to-speech generator and throw in some random screengrabs from existing videos", and I realize that any knowledge without a chain of custody will be drowned out by noise soon.
@mhoye
Not sure if you followed the saga of the Peppa Pig videos, but was more or less exactly what you describe--"semantic mulch"--but perhaps even more disturbing:
https://www.theguardian.com/technology/2018/jun/17/peppa-pig-youtube-weird-algorithms-automated-content