The very first YouTube video, uploaded by one of YouTube’s founders before it was purchased by Google, is called “Me at the zoo”. It was uploaded about 21 years ago. It has accumulated almost 400 million views and over 10 million comments. This is NOT the most watched YT video, which is a children’s video that doesn’t earn the creator money, with over 16 BILLION views. But that first YT video, only 19 seconds run time, noting how long the elephant trunks are, is still perhaps the quintessential example of what a good YouTube video SHOULD be, a slice of reality, a piece of true emotion.

Of course, that’s decades ago, and since then YouTube has become a massive money making machine for Google, and the quality of videos overall has plummeted to the Earth’s core. Because for Google, a click is a click, a view is a view, and Google cares little about what kind of content — other than blatantly illegal — generates those views of associated ads that now bracket and interrupt YouTube videos for non-paying viewers.

But it’s still the most important streaming platform on the planet. The technical infrastructure that supports YouTube is staggering in its size and complexity. And it’s not that there aren’t still wonderful, incredibly great videos on YouTube. There are, if you can find them. And that’s become increasingly difficult as AI-generated Slop increasingly permeates the platform, burying the good stuff under mountains of muck.

The relatively recent availability of powerful AI video generation tools has created a cosmic-sized mess. There are many types and trends in these. You almost certainly have seen them if you ever view YouTube. The previously rather robotic voices have vastly improved, but are (at least currently) still often recognizable. The happy, happy voice. The deep booming dignified voice. On and on. The quickie AI videos are churned out as rapidly and in as high volume as possible by those channels from automated text prompts since viewership of any individual video may be relatively limited.

They often have very recognizable characteristics. Slide shows with the images slowly moving around the frame. Mixes of images where some are utterly unrelated to the ostensible topic. Voices that pause in odd places or get obvious pronunciations wrong. Completely distorted images of objects or people from AI hallucinations. Lately new trends include AI videos ostensibly showing famous people saying or doing things that they may or may not have actually ever said or done. A famous deceased scientist supposedly narrating a text he wrote many years ago, sometimes including still or moving AI imagery of the scientist. Lots of images with annoying bits of what look like dirt or feathers or other debris constantly floating around over them. False or distorted histories, fake “revenge” or “confession” stories about evil Homeowner Associations, or awful bosses, or terrible spouses or other relatives. Endless numbers of such videos with minor variations. In the comments, usually some viewers are apparently taken in — “Oh, what an inspiring story!” — while other comments point out the AI Slop in very specific ways.

Keep in mind that YouTube does not reliably label as “synthetic or altered content”, even briefly, many AI-generated videos. By the way, my personal policy is to thumbs-down ANY video I run across that obviously is AI Slop.

But perhaps Google would be just fine with pretty much EVERY video on YouTube being created by Google AI — hey, no more creators to pay! After all, if viewers are willing to tolerate AI Slop, why bother presenting them with anything else? The YouTube money machine is heading toward a closed ecosystem — AI Slop flows out from YouTube, ad revenue flows in, with no room for genuine creators. YouTube may ultimately consider everything else to be expendable. I hope that doesn’t happen. But unfortunately, it’s not impossible either.

–Lauren–