The debates surrounding the increase in the presence of content generated by artificial intelligence take on a new dimension. Recent analyzes show a notable proliferation of this content, particularly on platforms such as Medium.
An unexpected prevalence of AI-generated articles
According to some studies, approximately 7% of articles published on news sites are generated by AI. However, this figure is much higher on some specific sites like Medium, where it exceeds 40%. This difference raises questions about the authenticity and quality of information available online.
Companies specializing in detecting AI-generated content, such as Pangram Labs and Originality AI, have carried out in-depth analyses. These companies have noted that a significant number of articles dealing with various subjects such as cryptoassets, marketing, SEO, or even technology, are produced by algorithms. For example, among the articles tagged “NFT” on Medium, almost 78% are suspected to be written by artificial intelligence.
The differentiated impact according to the themes addressed
It is interesting to note that the use of AI seems to vary considerably depending on the topics covered. In the beauty industry, many sponsored articles use outsourced writers using AI to generate content. On the other hand, technological and financial articles also seem to be strongly affected by this trend. There Techfinance, and science are sectors where we observe a lot of manipulated content, often linked to scams linked to crypto-assets.
According to McKenzie Sadeghi of NewsGuard, a company monitoring online misinformation, AI-generated content primarily consists of topics related to cryptoassets, marketing and SEO. This reveals opportunistic behavior aimed at attracting readership through automated and potentially unscrupulous means.
Reactions and measures taken by the platforms
Tony Stubblebine, CEO of Medium, denies the importance of the data presented by these detection companies. He argues that the actual impact of AI-generated content is minimal thanks to a combination of spam filters and human moderation. However, analyzes reveal that many articles suspected of being generated by AI receive several “claps” (equivalent to “likes”), indicating a notable audience in some cases.
Stubblebine also mentions the existence of accounts massively posting content without much interaction, suggesting a negligible impact. The idea that this content doesn't really affect the experiences of Medium readers is open to debate, particularly when the articles in question touch on widely followed topics like cryptocurrencies.
Global trends and international reactions
By analyzing on a larger scale, other surprising discoveries emerge. In a study of 26,675 news websites, start-up Pangram found that nearly 7% of daily articles were partially or entirely generated by AI. A significant portion of this content comes from countries like Ghana, which leads with almost 33% of AI-generated articles. Next come Peru, Brazil, Colombia and Pakistan.
These figures highlight not only a technical problem but also a geopolitical one. Some countries appear to be adopting these technologies more intensively than others, potentially influencing the overall media landscape.
Using AI detection tools: advantages and limitations
THE AI detection tools are improving, but they are not infallible. Jon Gillham, CEO of Originality AI, notes that while these tools can identify growing trends in AI writing, their accuracy is not absolute. They struggle to distinguish between text completely generated by AI and one only assisted by artificial intelligence tools, making it difficult to draw definitive conclusions on each individual content.
Nevertheless, these detectors are of undeniable importance for identifying changes in digital editorial practices. They help researchers, journalists and the general public to identify and understand the new dynamics at play on different online platforms.
The Skyrocketing AI-generated content requires real content creators to adjust their strategies. Risks exist regarding the dilution of the quality and originality of the content available. To preserve the integrity of their posts, platforms will need to redouble their efforts to combine effective automated detection with active human monitoring.
This context underlines the importance of mature and thoughtful regulation to balance technological innovation and journalistic ethics. Maintaining high standards of verification and authenticity is essential to maintaining public trust in online media.
