Google is aggressively hurting automotive content

For the past few years of my life, I have become obsessed with learning the YouTube algorithm. As the person ultimately calling the shots at the EV Pulse YouTube channel, it was my job to know what would work and what wouldn’t work.

Learning about that is a process. Some of that process is trial and error. Other parts of that process are diving deep into what YouTube says works and doesn’t work. Ultimately, it comes down to this: If you make a good video, people will watch it.

The algorithm doesn’t favor small channels or big channels. The algorithm is only concerned with the user’s watch time. Stuff that keeps the user on the platform, and thus makes YouTube more money through advertising, is stuff that will continue to be the most promoted.

How certain content is moderated, and what’s allowed on the platform, are different areas of discussion. But at the end of the day, if you make sure people want to watch, they’ll watch it.

At least, seemingly, up until recently. While YouTube initially shunned the idea of AI-generated content, it now allows it on the platform in many cases. More interestingly, and troublingly, the platform tries to determine your age based on an AI-based algorithm, and can enable a setting restricting some of the content you see because it thinks you’re too young.

This is different than the YouTube for Kids app, and the changes that come with marking a video suitable for children. As a side note, most creators never want to check the suitable for children box. It disables comments, restricts advertisers, and is much more restrictive on what you can post (even if you think your content is perfectly normal for someone under 13 years of age).

But I digress.

Views are down across a number of popular channels, including some that I watch on the regular. Some people blame the new AI-based age filtering; others claim it’s just YouTube cleaning up bot accounts and that revenue hasn’t changed. The Spiffing Brit, whose entire schtick is breaking stuff (including YouTube), says it’s just the algorithm recommending good content, and that the current drop is a seasonal thing that happens every fall.

I tend to side with Spiff on this, as people are chronically online during the summer months, but then head back to school in the fall and drop their YouTube viewing. But with so many creators complaining, it does feel like there’s more to it than just a seasonal difference.

And that brings me to Google’s search product. Many publishers are reporting a significant drop in search traffic across their properties, and the blame lies squarely with AI-based search.

Some reports indicate that only a very small percentage of people using Google search these days actually click the link. Google has almost always farmed website content to provide results without a click, but the AI-based feature makes matters even worse. You’d think that a search engine would want to drive traffic to sites to find answers, but Google seemingly wants you to find the answer only on Google, and not leave the platform.

As a result, many big publications are leaning more and more into AI themselves for content generation. With declining revenues, AI provides a way to eliminate staff positions while still generating a continuous flow of content.

I’m not here to judge whether or not you use AI, or how you use it. I am here to judge Google’s use of it, and how it’s ultimately self-defeating for Google to continue to use it to drive search.

As you know, Google’s AI has to scrape the web for information so it can respond to search queries. When that content is created by intelligent people writing well-researched analysis, then the results Google returns might be pretty decent (though your mileage most certainly will vary).

But as more and more content created comes from AI, and less from human sources, the entire thing gets worse. The snake eats its tail, and results are generated from content that itself was generated from content that was also generated by AI. The large language model (LLM) doesn’t get more “intelligent,” it gets stupider.

It doesn’t just get stupider for Google, though. It gets stupider for everyone using AI (even for non-nefarious reasons) because it’s not learning from original content; it’s learning from generated content.

This means that the value that Google offers to people is less and less. If YouTube’s creators aren’t provided the motivation they need to keep creating on the platform, some will leave. It’s hard to say that any YouTube competitor could take off at this point, but the rush to AI might just be the catalyst a competitor could use to grow.

If you are using AI for some mundane tasks, like updating automobile buying guides or generating vehicle overviews, over time, the information will be less and less reliable, meaning that a human is going to have to spend more and more time reviewing and editing the content for accuracy.

What I do know is that many automakers are leaning into AI more and more. With CoPilot coming with Microsoft 365, it’s something that can be integrated into the workflows of many of these companies, and with advanced corporate management functions, IT departments can track how you use AI and ensure that you are using it for your tasks.

There is a time and a place for AI, and there is a time and a place for LLMs specifically. But while Google hurts publishers and content creators to raise the value of its stock, this really does feel like the 1990s dot-com bubble, and when this bubble bursts, Google is going to be in a world of hurt.