When we think of AI, the first major benefit for business and society is task automation, which frees people from routine work so they can focus, in theory, on more creative work.
The arrival of generative artificial intelligence is changing this perception: will AI replace us in creative work too?
In practice, today’s new algorithms enable us to create content, images, music and videos.
With the arrival of ChatGPT, the whole world seems to have found the Holy Grail of content marketing: we can quickly create and publish lots of content online.
I do not deny that, as a Content Strategy enthusiast, I play a lot with generative AI to understand where it fits in the creative process.
The first question that naturally arises is:
if we produce our online content with ChatGPT (or similar tools), how will search engines treat it?
This is a fundamental question for Content Marketing.
Can Google recognise AI-produced content?
It is difficult to answer. On the one hand, the Mountain View company is very advanced in artificial intelligence and content analysis — after all, that is its business. On the other, a few changes and another pass over the text can make its origin almost impossible to recognise.
Personally, I would not challenge Google by publishing large amounts of content exactly as ChatGPT churns it out.
There are various content-detection tools around, including professional ones, that are supposed to tell us whether content was created with artificial intelligence.
The tests I carried out gave poor results: even famous literary works were judged to have been produced by AI!
Yet the creative provenance of content is a fundamental issue. As with the food we buy, we need to know the ingredients and where they come from.
One solution comes from OpenAI itself, which could add cryptographic watermarking to identify content produced with ChatGPT and all its APIs.
For transparency, content producers themselves could state whether they used an AI tool.
But what does Google say?
We know that the famous search engine normally provides guidelines specifically to help content creators.
Here is what it says:
‘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.’
So Google tells us that, if we use AI to manipulate rankings, we will be treated as spammers and our content will never see the light of day in its search results.
For Google, manipulation means automatically creating new content with no added value, solely to cover topics and appear in results.
But note that Google also says:
‘Appropriate use of AI or automation is not against our guidelines. Provided it is not used to generate content primarily to manipulate search rankings, which is against our spam policies.’
In essence, if you want to use ChatGPT and AI to create content cheaply, quickly and easily and throw it onto the web, Google will not love you and you will be considered a spammer.
But it also means using AI is fine and Google is happy if you use it to help build valuable, useful and original content (according to Google’s guidelines).
Because we know the search engine, by its nature as a broker, is hungry for valuable material: its success depends precisely on its ability to find it and present it to users.
With that said, what might be a smart way to use tools such as ChatGPT for content marketing and SEO?
One possible approach is as follows:
- Analyse the needs we want our content to meet; identify the questions, useful topics and most frequent searches of our audience. We can use various tools for this, starting with Google Trends (user guide) and other tools such as Semrush and Ubersuggest (user guide).
- The output of that analysis is a series of questions, words and concepts. In marketing, for example, these might be: what marketing is, how to create a marketing plan, which tools to use, which skills are required, what the marketing mix is, etc.
- At this point, we can create a prompt for ChatGPT saying: ‘Write an article outline with a title and subheadings that answer or contain the following questions or keywords: [put our list here].’
In doing so, we harness the strength of language models: their ability to organise and catalogue knowledge.
Once we have the structure, we develop the content of each section. If we do not know how to arrange the content, we can again ask ChatGPT for help, but only for ideas about structure: I do not recommend copying and pasting.