Until we reach a point where artificial intelligence is capable of actual thinking, the AI content will not be suitable for SEO. And there are mathematical reasons why any attempt to manipulate the algorithm will be short-lived.
This doesn’t mean AI content is not suitable for any use case. For example, this limitation is quite welcome in most business chatbot applications. But SEO requires specific attributes that are not logically (and mathematically) possible in today’s algorithms.
I will discuss in detail the inner workings and capabilities of AI:
- Why is AI content not a good fit for SEO
- Why some sectors welcome AI-generated content
- The future of AI content
Why Is AI Content Not A Good Fit For SEO?
The short answer is that it lacks originality and unique ideas. It simply cannot produce anything that was not seen before. And search engines aren’t going to put a piece of content on the first page of the results that has nothing new to offer.
So the question “can Google detect AI-generated content?” is, in this case, irrelevant. Because search engines can detect if the content doesn’t bring anything new to the table. And this is the case with all AI content and some of the manual content as well.
But why can’t AI content “bring anything new to the table“?
Difference Between Strong vs. Weak AI
AI is most impressive to people who never worked in this field. It’s just like a magic trick – it’s mind-blowing as long as you don’t know how it works.
You see, the real genius behind artificial intelligence is not inside the algorithms. Instead, it’s inside the minds of AI developers.
AI is like a giant hydroelectric power plant that lights up the entire city. It’s almost magical. But its components are super simple: mostly water, some concrete, and a little bit of metal. And that’s it.
The real genius is assembling those parts into a functioning power source.
Most people think that most AIs work like strong AI. Meaning AI is capable of thought and comes to its own conclusions. But exactly the opposite is true. None of the AI has these attributes.
And many people argue that AI will never achieve the capability of independent thought.
What is currently in the works is the weak AI. It’s an algorithm capable of solving very narrow tasks and nothing more. In most cases, AI is just taking the average of two numbers. Literally takes 1 and 4 and “realizes” that 2.5 is the result. This is the state-of-the-art AI we have today.
And that is why I say the real genius is in the AI developers who assemble these unbelievably simple algorithms to actually do something useful.
How Is AI Content Generated?
AI content cannot bring anything new to the table for SEO because it’s just taking averages (from words this time). For example, it can take best-ranking content from search results, mix it up, and spit some mixture derived from existing articles.
But all of this doesn’t bring any additional information or new content. And new content doesn’t mean spinning the same message using different words:
- “Today it’s sunny.”
- “The weather is nice today.”
If we just look at the words and letters that were used, those two are very different sentences. But the meaning is exactly the same. This means there is no new “content” in the second sentence that was not already communicated in the first one.
Mostly, this is how AI-generated content works. It just spits out the mixture of existing content and nothing else.
Besides, you wouldn’t want this AI to spit out new content because then it would have to invent new content. And since it doesn’t understand the content, it would just introduce new random (most likely false) information.
What Search Engines Want That AI Content Lacks?
Ideally, the first search result would contain all the relevant information the user is interested in. A person would read it, and all their questions would be answered.
In reality, sometimes it’s hard to even guess what the search intent is just from the keyword. Especially if it’s a broad keyword.
For example, if someone typed “bitcoin”, their intent might be to:
- Just see what the current price is
- Maybe to find out what bitcoin is
- Or see the latest news about it
And the point of the first page of search results is to contain all relevant information for a specific keyword.
This covers cases where the first search result does not contain all relevant info for the search intent. And then, the second search result contains information missing from the first search result and so on.
So even though none of the articles contain the complete info, the first page of search results might. And this is where the fundamental problem with AI content comes into play.
If this content is just shuffling the existing information from the first page, it brings no additional information (and no value). Even if it’s rephrased, mixed up, and textually “unique”.
The problem is that it contains zero extra information. That is, it doesn’t contribute to the conversation. It just regurgitates the existing information.
And because of that, this content is useless and will not be relevant to the search results (i.e., it will not be ranked).
Now I hope you understand why it’s not important whether search engines can recognize AI content or not. They can recognize irrelevant content, and that’s enough.
Why Some Sectors Welcome AI-Generated Content
Some use cases either don’t require or specifically prohibit original thought. For example, let’s say you had an AI cake recipe generator. You wouldn’t want it to have an original idea of swapping sugar with salt.
So in certain situations, you want “novelty”, which is better defined as variety. You want this AI to operate inside of the existing area of possibilities. And you want it to find variations that were previously missed. But you do not want it outside of the predefined realm of possibilities.
One such area is customer support or, more specifically, support chatbots. Whatever the customer asks, you want to provide an answer that you previously approved in some way.
Since it would be very bad if the chatbot started giving out “creative” massive discounts you haven’t approved previously. Or any other answer you might disagree with.
And this is the area where AI still can (and should improve). But it is perfect for weak AI because chatbots’ only task is to regurgitate existing info at the appropriate time (i.e., when the users ask them to).
The biggest challenge here is how to classify/identify what exactly the customer asked and pair it with existing answers. All in all, I see a bright future for customer support chatbots.
Video thumbnails don’t seem to have much in common with SEO at first glance. But more engaging thumbnails influence the click-through rate. And this impacts search engines’ interpretation of the content quality.
This is also a good area for current AI to tackle.
There are thousands of parameters that AI can analyze, see what successful thumbnails have in common, and produce something similar:
- Which colors were used
- Was there any text, and how much
- Which words were used
- Are there any images of faces, people, or animals
- What facial expressions were the most successful (happy, sad, shocked)
- What were the positions and sizes of certain elements
- What type of thumbnails was the best for the target demographic (age, country, interests)
As you can see, this is a perfect task for AI because it can take into account thousands of features that a regular person simply cannot. And if it’s good, it can do a better job than a person.
So using AI-generated thumbnails is a great way to boost your SEO rankings.
In theory, this should work very well. But currently, in practice, I know it doesn’t work. How do I know this? Because companies who make these tools hire copywriters to write their sales pages.
So if they’re not willing to use their own copywriting for their business, neither am I.
But over time, this should work great. It is “easy” for AI since sales don’t change much over time (because psychology doesn’t change over time).
People want the same things: money, food, beauty, attention, easy life, etc. – It’s a constant. And using words that imply or promise these things requires nothing innovative (from the info/message side).
Besides, once you go a little bit deeper into copywriting, you realize there are “formulas”. And the more you stick to them, the better your copy is and the better results you get.
So AI copywriting content should definitely prove successful sooner or later.
The Future Of AI Content
AI is certainly going to advance (and a lot). But its advancement will mostly depend (and be limited by) computing power and “clever programming”. And I would not expect any major breakthroughs, just better ways of assembling the same simple algorithms we already have today.
Computing power helps.
When you look at the “progress” of chess AI, you may think there was some breakthrough and AI is unbeatable. But put any algorithm from today in the computer from the 1980s, and you’ll see they are still horrible.
I’m not saying this to bash AI. Quite the contrary.
It is almost guaranteed that computing power will continue growing, and that means great news for all tasks that are well defined inside narrow parameters. All these tasks/problems will be solved using the power of AI (purely through brute computer force).
So the majority of AI’s advances will come through using our brain power and finding ways to reduce the complexity of the tasks so they can be picked up by AI.
In general, more computing power means that worse algorithms will appear to work in some narrow cases. But always remember, the world is much more “messy” than we realize.
Personally, I fail to see any progress towards any kind of original idea or “thought” that can be generated with AI. Sure, many demonstrations look impressive, but as soon as you scratch the surface, you realize it’s another gimmick.
It always boils down to reducing the complexity and increasing the computing power. For example, suppose you saw AI play some computer game very well. But then you realize it can win only in extremely narrow situations (only certain maps and certain moves are allowed).
And at the same time, it requires over 120.000 processors to train the AI model. Some may say this is progress, but realistically, there is no progress if you have to simplify the game to reduce it to chess. On the other hand, you throw unimaginable amounts of computing power.
What about AI-generated art?
Again, it is just shuffling the existing art around (i.e., its training data). But that’s just part of it. The second part of AI is in your mind. Let me explain…
Have you ever looked at your dog (or a cat) and felt like they understand you? Just from the way they look at you?
In reality, they may sense something, but it’s far from understanding. Nonetheless, you feel understood. But the understanding was provided by you, not the animal itself.
And in the same way, AI-generated art can be much the same way. Whatever image you see, your mind will provide an interpretation. So it creates an illusion of creativity.
In fact, it is simpler than text-generated content. Because AI can throw a bunch of random words, you will notice it makes no sense. But if it throws a bunch of random colors and objects on an image, we interpret it as art.
To write something unique, you need to think of something unique. And as long as AI is not at the level of independent thinking, it will always just have to reiterate existing content (if it wants to be useful).
Personally, I am happy that AI is going to replace us in repetitive and boring tasks. Because that will free up our creative energy to be used in areas that really matter. But, whatever you do, don’t use it for SEO.
Senior Software Developer with experience in Machine Learning, Artificial Intelligence and search engine algorithms.