This seemed to come as a bit of a shock to some, but I think there are three reasons why it makes total sense.
Google has many search algorithms working behind the scenes to rank the best results for its users. Are they perfect? No. But I think we can all agree they do a good job on the whole.
For example, there are 717 million results for “lump on neck—yet none of the top ones are from your average Joe. They’re all written by doctors and published on the sites of respected health brands. In other words, they have E-E-A-T.
Of course, E-E-A-T isn’t a direct ranking factor, so let’s look at a couple of known direct ranking factors and discuss a couple of reasons why, on the whole, AI content won’t flummox them.
People link to content for all kinds of reasons, but I think these are the two most common:
- They already know and trust the author or brand.
- The content is the source of a unique idea or statistic.
Given that it takes years to earn trust in an industry (even if you only publish high-quality content), it’s not going to happen if you only publish low-end AI content.
As for content that says something unique, take a look at these “unique” link building tactics ChatGPT came up with:
If you thought you heard about these 20 years ago, you’d be right. They’re not unique ideas at all. You’re going to have to come up with these on your own.
Helpful content system
Google’s helpful content system aims to boost the performance of content that gives visitors a satisfying experience while demoting content that doesn’t.
Here’s how Google says it works in a nutshell:
The system generates a site-wide signal that we consider among many other signals for ranking web pages. Our systems automatically identify content that seems to have little value, low-added value or is otherwise not particularly helpful to those doing searches.
Given that AI tools are trained on existing content, it’s basically impossible for them to add value. They can only summarize and rework ideas that are already out there (or that you give them).
Google also states that this is a sitewide signal. This means it should still negatively impact those who think, Meh. I’ll just publish a load of AI content, see what sticks, then improve it.
Google couldn’t be any clearer about this. Its documentation includes an entire section about how automation can help create helpful content.
Here’s an excerpt:
Not all use of automation, including AI generation, is spam. […] AI has the ability to power new levels of expression and creativity, and to serve as a critical tool to help people create great content for the web.
If you’re wondering how, let me share three examples.
AI can make your content interactive
Let’s say you wanted to rank in the U.K. for UK tax brackets. If we look at the top results, they’re all about the same. They list tax brackets and explain how to determine which one you’re in.
This is useful information, but you’ll struggle to calculate your personal tax liability in your head. You need a calculator, and it would probably be helpful if it was embedded right there in the post.
Before AI, unless you knew how to code, you’d have to hire a developer to do this for you. But now, with ChatGPT (and presumably other upcoming AI tools, like Google’s Bard), all you have to do is ask:
Here’s the output:
The calculator doesn’t account for the tapered annual allowance that (currently) applies to income over £240K in the U.K. But you could probably get ChatGPT to add this with a few extra prompts.
AI can proofread stuff
Last year, we tested the quality of freelance writers based on their rates. The cheapest charged just $0.02 per word.
If we take a part of their content and ask ChatGPT to proofread it, it makes minor improvements for clarity:
I’ve found this super helpful for improving spelling and grammatical issues in my drafts. Here it is correcting all of my typos and even capitalizing acronyms without me having to lift a finger:
AI can explain things better (and faster) than you
Here’s what happened when I asked ChatGPT to explain and name a budgeting hack:
I don’t know about you, but I think it explained my admittedly lame (and somewhat unrealistic in this economic climate) hack pretty well. Sure, the name isn’t that great, but we can always ask ChatGPT for some other ideas:
“Round Up Retreat” seems pretty catchy to me.
No. Bad AI content will rank for some keywords, just as low-quality content written by humans does. Here are two reasons why.
Google isn’t perfect
I’m sure some of you rolled your eyes when I mentioned backlinks and “helpful content.” After all, you can always buy backlinks (although we don’t recommend it), and there are plenty of sites with unhelpful content still thriving.
This is, in part, because Google’s algorithms aren’t perfect.
You can think of Google Search as one of those toys with holes that only let through specific shapes. They work, but if you spend some time twisting and turning the shapes (or grab a hammer), you can often force the wrong ones through.
Just like these toys, Google’s algorithms aren’t going to catch all low-quality content—AI or not.
Google can only rank good content if it exists
Google’s algorithms are designed to surface the best results, but you’d be forgiven for thinking otherwise based on some SERPs.
For example, here’s the top result for “best headphones for big heads”:
You can see that it’s bog-standard affiliate content. Lots of words, no unique insights, and stock images. In other words, the level of quality that AI tools could easily write.
So how can this possibly be the best result for the topic?
I think the answer is simple. The topic just isn’t lucrative enough to incentivize meaningful competition.
For example, the top-ranking page only gets an estimated 430 monthly search visits:
Let’s be (very) generous and assume that:
- Each recommended pair of headphones costs $200 on average.
- 10% of visits lead to purchases.
- The site gets 3% commissions from Amazon.
Do the math: $200 * 43 (10% of 430 visits) * 3% = $258/month.
This isn’t bad, but I still doubt anyone would be willing to buy and test dozens of $200 headphones to create exceptional content for this keyword. This creates a void of truly useful results for Google to choose from and a low bar for the best result.
In fact, ChatGPT could probably create content as good as the current top result.
Marie Haynes summarizes Google’s stance on AI content well in this tweet:
Google’s taking this stance because it knows AI tools are like power tools. If you’re a carpenter, they’ll help you get the shelves up faster and with a better finish. If not, they’ll just help you botch the job faster—and take one of your fingers in the process.
But while some SEOs will lose a metaphorical finger (or two) by jumping knee-deep into AI content, others will make it work. Google’s algorithms are good, not perfect. So expect to see a few articles on shoddy AI content sites ranking like crazy and making bank.
Inevitably, though, these sites will tank. Not because they published AI content, but because they published low-quality content that never deserved to rank anyway.
Got questions? Disagree with me? Ping me on Twitter.