Whether it’s doing homework, summarizing a book or writing a blog article, there’s not much that generative AI tools such as ChatGPT can’t do. Large language models (LLM) can even write SEO texts for you. Just enter a few words (or ask the engine to suggest some) and they will be reliably inserted into the text in the right places. And it seems to produce good rankings.
“Yes, it works,” concludes Joahnnes Beus, SEO Expert at analysis company Sistrix, after a test was carried out with the US finance website bankrate.com.
The discussion really got going after it became known that the tech news website CNET uses AI to write some of its articles – and achieves very reasonable ratings with it. Furthermore, the news outlet admitted that a human gives the articles a few finishing touches. The editorial team monitors and edits the texts.
However, it’s not quite so simple. Let’s take a closer look.
What does Google think about AI texts?
Danny Sullivan, SEO Expert at Google, explained the company’s official stance on the Google Blog in February:
“We focus on the quality of content, rather than how content is produced.”
Consequently, a few days ago, the SEO Playbook was updated, and the “Content written by people, for people” section has been dropped.
This means that Google isn’t interested in who or what writes the content. If the content is ranked as being valuable to users, Google ranks it well. Google explains what exactly “valuable” means in its “Search Central” documentation.
A central aspect is the EEAT rules (experience, expertise, authoritativeness, trustworthiness). So website operators would do well to produce content that demonstrates experience, expertise and reliability. How exactly this is determined still remains a mystery.
Can Google even recognize AI texts?
Doubts have arisen, especially since the OpenAI Classifier tool was shut down. It became clear that even the company behind ChatGPT couldn’t reliably identify AI-generated content.
Google has also been developing a whole range of AI tools to better understand texts. The Natural Language Processing (NLP) Service can, for example, analyze texts for their meaning. It remains to be seen if tools like these would be able to recognize who or what has written a text.
Reliably detecting AI texts is an almost insolvable problem, as there are now umpteen different models that are all specialized in producing natural sounding content – and they are constantly being refined. It’s a race which even Google can hardly win in the long-term.
What does this mean for content creators?
There is nothing wrong with using AI tools. However, when it comes to SEO, content with real added value wins out – for people. There’s a good reason why the dwell time is still one of the most important ranking criteria.
AI can help support the thought processes involved in creating added value, but it can’t replace them, as with copywriting, it’s more about thinking, rather than writing. Knowing exactly what the finished text should be about remains the greatest challenge: it’s called prompt engineering.
Expertise and industry knowledge can’t be automatically generated, and neither can curiosity and creativity. When it comes to thinking outside the box, experimenting and being surprising, AI can help, but it isn’t the central success factor.
The result? The decisive role is still played by people.
We predict that people will still be involved in creating SEO content in the future. But this doesn’t mean that ChatGPT and other tools should be ignored. Quite the opposite. Combining human ideas with the efficiency of AI is the path to success. And it’s one we’re happy to accompany you on.
Cover image via Pexels