It is increasingly difficult to distinguish content written using generative artificial intelligence, since it is progressively being integrated at all levels, especially on Linkedin. According to a study released by Wired and conducted by Originality AI, it appears that 54% of the longest posts on the platform were “probably” generated by AI.
AI content increased by 189%
For this study, 8,795 publications distributed between January 2018 and October 2024, containing more than 100 words, were analyzed. Although the first years are not very representative of the current phenomenon, the year 2023 marked a turning point. “There was a spike when ChatGPT came out,” says Jon Gillham, director of Originality AI.
This increase amounts to 189%, although a stabilization has since been observed.
This study is released as LinkedIn encourages the widespread use of AI through its Linkedin Premium offering. Subscribers have access to internal tools to “rewrite” posts, profiles, CVs and even private messages.
Linkedin defended itself by claiming to have several methods to identify and ban “duplicate or poor quality” content in users’ feeds, while promoting the adoption of AI on the Microsoft-owned platform.
In testimonials collected by Wired, users explain that they use AI to write “post drafts”, which they then modify considerably. One content creator for several companies said it saved her time. However, it relies on large language models rather than specialized tools, notably Claude, developed by Anthropic.
AI is also used to post messages in English by those who do not have a good command of the language. However, the popularity of some AI-generated posts raises questions about the quality of users’ feeds and poses risks of abuse.
Source: BFM TV
