Most Long LinkedIn Posts Are Likely AI-Generated, Study Finds

A recent study by AI-detection service Originality.ai found that 54% of LinkedIn posts with 100 words or more are likely to be AI-generated. The study, which analyzed 8,000 posts over 82 months, also found an increase of 107% in the average length per post.

The study classifies posts as AI-generated if Originality.ai’s tool flagged them as entirely generated by AI and unedited, generated by AI but human-edited, or human-written but heavily AI-edited. Between January and February 2023, just a couple of months after ChatGPT’s initial public launch, the number of AI-generated posts on LinkedIn increased by 186%, and that higher rate has held steady.

Earlier this year, LinkedIn started rolling out premium company page subscriptions. Among other features, this subscription gives users various tools for generating AI content. However, the increase in AI-generated posts on the platform started much sooner than that announcement. It aligns with the public release of ChatGPT and later with the release of its GPT-4 generative model.

LinkedIn has seen an average increase of 20 million registered users per quarter between 2020’s first and 2024’s third quarters, with around 990 million members at the time of writing. This explosion in popularity makes it a ripe space for content creation, including AI-generated posts.

It’s important to note that AI detection tools aren’t perfect, and it’s easy to circumvent them. Recently, OpenAI, ChatGPT’s parent company, had to shut down its own AI-detection tool due to low accuracy results.

However, a meta-analysis of third-party studies conducted by Originality.ai found that the service was the most accurate AI-detection tool in six studies and among the three most accurate in the remaining two, lending credence to its findings.