Barnaba's guidance on the content generated by AI
The topic is quite perverse and I will try to describe it in more than just a subjective way.
Artificial intelligence is the word of the year. In articles,posts it was inflected with all forms and cases. The digital marketing consultant explains:
How artificial intelligence is used and associated
When AI, Chat GPT was more widely discussed in the media two main issues appeared. The use of Chat GPT for software development, and for building content. From the beginning, AI used for software development, created positive wonder. The programmes work, the code is simple, fast. In a word - a super solution. AI in the aspect of graphics creation received similarly positive reviews. The world went crazy under the impression of what AI was generating. Everything seemed to take off and there would be no end to the successes. The processes involved in generating long-form content proved problematic. I have thrown this topic into the spotlight.
My first and second experiences with AI.
My first experiences with Chat GPT were painful to say the least. I tried to generate a blog post about the Bieszczady Mountains. Stylistic errors, stuffing of phrases. The texts turned out to be a disaster. The first version of GPT Chat as a text generator I pushed out of my workshop.
The generation of graphics is completely different. Here, a very interesting tool was made available by Bing. Here, the case required a few adjustments to the queries. I precisely defined four features of the graphic. I obtained a very promising input. I was able to combine the generated graphics into one and use it as a thumbnail for my Youtube videos. A very useful tool.
How I use AI for my own work
I currently use Chat GPT-4 for my work. The world has moved on, but not AI. At least not to the extent that it needs to. Still, the generation of content that is supposed to be long is of very low quality. An order specifying a particular phrase, the length of the text, results in content with repetitive-sounding paragraphs. In a word - garbage.
After these devastating reviews of Chat GPT as a content generator - it is important to recognise that the tool is not completely useless. Chat does a very good job of generating short content. For example, I gain invaluable support from AI in generating short paragraphs with a specific phrase. I use these paragraphs to add internal linking at the end of product descriptions. Preparing 30 or 90 pieces of content of this type is absolutely no problem. Inserting a link to a category page from a specific phrase into the prepared content is a formality. In moments, I have a job done which, if taken in creative form, would have taken a long time. In tests conducted over several days, I have determined that Chat GPT-4 is most efficient with texts up to around 2,000 characters with spaces.
What is Google's official position
Here the paragraph could be simplified to a quote from Google's documentation
When it comes to automatically generated content, our guidance has been consistent for years. Using automation—including AI—to generate content with the primary purpose of manipulating ranking in search results is a violation of our spam policies.
To make readers aware of the seriousness of the issue, it is important to inform them that long before Chat GPT was published, there were algorithms for evaluating content. A few years ago, BERT, an algorithm using machine learning to understand content literally, was implemented. BERT has been quietly and intensively learning how to recognise written content in a way that is consistent and coherent with the quality of the author. A separate thread opens with the relatively fresh EAT algorithm, which evolved into EEAT. These algorithms judge the quality of content in terms of its usefulness to the reader and the authenticity of the author's experience.
I have a firm belief, which comes from experience - that the above does not convince many businesses. Finally, I present a case study - where Chat GPT was used to manipulate content.