Using AI to generate blog articles: not such a good idea

As the use of AI continues to grow, many businesses are using this technology to simplify the creation of content and more specifically for writing blog articles. 48% of marketing professionals¹ admit to using AI for this reason.

However, AI can reduce the quality of content and be detrimental to your rankings in search engine results pages (SERP).

Read how Google handles AI generated articles, the difference between AI and human written texts, and how AI can be useful for writing product datasheets.

AI generated blog articles: what Google has to say

Google: the importance of quality

Google's ranking systems put more importance on the quality¹ of content than on how it was produced.

To assess the quality of content, Google uses the E-E-A-T criteria which stands for Expertise, Experience, Authoritativeness and Trustworthiness. These criteria enable Google to ensure that the content available to Internet users is « original, high-quality and created to benefit people ».

Google's position on AI

Google considers that an appropriate use of AI does not « conflict with its guidelines² », and recognizes the value of AI in some situations: sports results, weather forecasts, and transcriptions. However, if a business generates AI to influence its website rankings, this goes against « Google's spam rules² ». Google uses various systems such as « SpamBrain² » to detect « the same patterns and signals » within a text to identify its content as spam.

Guidelines on AI

According to Google, if you use AI to create content, this must be indicated when the origins of the content produced « is not implicit ». Google's position on this, which is not very clear, raises a question of ethics: should we systematically indicate the use (or not) of AI to produce content. The web giant has also taken concrete steps concerning images, by equipping itself with « SynthID ». This tool places a « digital tattoo³ » on all images generated by Google Imagen.

Google also announced the extension of SynthID capabilitiese⁴ to texts generated in the Web Gemini application, and to Veo videos.

Writing blog articles: the difference between humans and AI

The risks of excessive use of AI in producing blog articles

• Risk of duplicate content: AI may « paraphrase⁵ » information on which it has been trained and may also generate texts that are too similar to existing content⁶.
• Risk of diminished quality: due to the excessive use of AI, standardized content is increasing and spamming⁶ the Internet: same structure, same ideas, same writing style… With the multiplication of this content comes the risk of diminishing the overall quality of the resources available on the Internet.
• Risk of sharing fake information: some AI tools such as ChatGPT have a tendency to « hallucinate⁷ » and generate incorrect information and/or quotations leaving 66% of Internet users worried about this⁸.
• Risk of reduced SEO performance: very often, businesses that delegate the generation of their blog articles to AI see a reduction in their SEO traffic⁹. The positioning and traffic stats of their websites drop significantly after a few weeks or months.

The value of the human mind in writing articles

Several elements differentiate articles produced by humans from those produced by AI:

• Analytical ability: a human writer goes further than AI in their search for information and is capable of simultaneously mobilizing many different abilities during their search: analytical ability, questioning the reliability of sources, evaluation of consistency… therefore enabling them to create more reliable texts.
• Authenticity: while AI follows precise instructions or prompts, a human writer is capable of distancing themselves from the indications provided initially so that while they are writing they can constantly question and adjust their texts: Is this globally consistent? Which writing style should I use to make my brand stand out? How should I word this sentence to get my idea across? How can I moderate my thoughts?...
• Expertise: The Internet does not have all knowledge on a topic, AI is limited to the information available on the Internet to generate its texts. A human writer has no limits and can supplement their thoughts on a topic using other resources: on field experience, interviews with employees, customer feedback… This provides them with additional information and allows them to look at a situation from a new perspective.

AI and text generation: the specific case of product datasheets

It is not recommended to use AI for creating texts with high added value. However it can be used to produce large quantities of a certain type of content such as product datasheets.

Contrary to blog articles, product datasheets do not use data from the Internet, but rather the product information available from various sources within the business: suppliers, distributors, manufacturers… AI allows you to use this information to create large numbers of product datasheets in record time: AI can reduce the time spent on this task by 80%¹⁰. However, human input is still required to validate your datasheets before publication.

Some Product Information Management solutions have integrated AI functionalities to facilitate the creation of product datasheets. With PIM, businesses can save a considerable amount of time in the production of product datasheets, simplify the data enrichment process, and guarantee the reliability of the information presented.

¹ HubSpot. Les statistiques IA indispensables à connaître en 2024. hubspot.fr.
² Conseils de la recherche Google concernant le contenu généré par IA. (2023, February 8). developers.google.com.
³ Defer, A. (2023, August 30). DeepMind dévoile SynthID, un outil permettant d'identifier les images générées par Google Imagen. usine-digitale.fr.
⁴ Semrush Team. (2024, May 14). Comment détecter les contenus rédigés par l'IA et le plagiat ? semrush.com.
⁵ Sebag, R. (2023, November 9). Génération automatique de contenu : l'IA peut-elle remplacer les rédacteurs SEO ? journaldunet.com.
⁶ Jonniaux, A. (2022, December 26). Les textes générés par IA polluent déjà l'Internet mondial. Journal Du Geek.
⁷ Seramour, C. (2023, June 9). OpenAI poursuivi pour les hallucinations de ChatGPT. lemondeinformatique.fr.
⁸ Gaudiaut, T. (2024, April 30). Quelles sont les plus grandes menaces liées à l'IA générative ? Statista.
⁹ Giansetto, A. (2023, November 9). Rédiger ses contenus SEO avec l'IA : bonne ou mauvaise idée ? Payplug.
¹⁰ Eugène, M. (2023, November 22). Comment booster vos fiches produits avec l'IA générative. BDM.