The popular SEO plugin Rank Math plugin has added in its latest update an artificial intelligence-based content wizard called Content AI.
They claim that this assistant uses its own artificial intelligence engine, which “learns” from data from various sources, although there is no information about these sources.
I insist, it is an assistant. I repeat this because there are many people who, on reading “artificial intelligence” and “content”, have been quick to get excited, believing that this is a dumb machine that will generate quality articles at the click of a button.
As with almost all similar AI services, what you get are various suggestions on articles already written
The suggestions are mainly keywords, titles, questions, tags, headline tag distribution and related links to supposedly “push” your rankings in the search results by attacking those terms that are most searched for or being searched for at a given time.
Although I’m not a big fan of these tools because they suggest texts and structures that turn texts into ridiculously unnatural pieces, I tried it since they were giving away some credits.
To start with, you have to activate the module and in “Settings” choose the language. You have a “Worldwide” option to make it work in all the languages it supports, of which there are about 80 at the moment.
The good thing about the functions being modular is that you can deactivate them if you don’t plan to use them or don’t need them temporarily
Anyway, they assure that this new module will hardly have any impact on performance as it only adds 7 files and 492 lines of code (which may increase or decrease in the future), and all the work is done on the Rank Math AI server. I haven’t noticed any noticeable overhead with the module enabled.
When Content AI is activated, when you open the editor, the access button appears.
The approach to the initial credit distribution is quite tight and I think it is a mistake
They offer 5 trial credits for the free version of the plugin, 25 for the PRO version, 200 for the Business version and 750 for the Agency version.
At the moment, if you want more credits you have to upgrade the version because you can’t buy “loose” credits, although they promise that soon you will be able to buy them without having to upgrade to the more expensive versions (which you probably don’t need and can’t afford to pay for).
Each word search means spending one credit and the data for this word is saved so that you can review it whenever you want.
The first thing you notice is that the tool is very green and still needs to be nourished with sufficient data and references
In addition to its own errors when faced with the ambiguity of certain words or phrases and unrelated results based on more general terms, there are suggestions in other languages or with a mixture of several languages and unrelated tag suggestions
In the suggestions of links to add to your article is where it fails the most, in the first five tests it failed in four. It suggested links to Instagram accounts or posts with no link to what I wrote and in the worst case it didn’t find any links at all.
Inconsistencies are also found in media suggestions (images and videos) header tags in long texts. Something similar happens with tags and keywords, proposing impossible or erroneous phrases in their construction. This is supposed to improve as staff use it. As with the suggestions and SEO metrics of the plugin itself, it is rational to use those of this “Content AI” module as a possible clue.
That is, a simple reference to improve your texts a bit more for search engines, never as goals to be achieved, unless you want your articles to look like they were written by a stuttering robot.
It works well and integrates well with Elementor. The look and layout is very reminiscent of the interface of SEO Surfer.
At the moment it is not getting very good reviews, from what I have been able to read in the facebook group rank Math. The strange scores and measurements in the analysis, bad results after spending credits and the current system to get them are the culprits.
Here you can check the user guide.