Substack’s new AI detection

Substack introduced a tool that detects “AI-assisted writing” in posts. Of course they did. Many of the most successful Substacks are run by professional writers, who are naturally concerned that AI could help others compete with their content. So creating a culture where using AI for writing is generally frowned upon is appealing to them.

Of course, using an AI-based tool to detect AI writing is the sloppiest way to fight AI slop. It does not consider the quality of the result, only the tools used. Useless content predates AI – it used to be called clickbait, often created by humans in so-called content farms.

It is better to continue using your own judgment about whether a piece is worth reading, rather than relying on a simple metric that tries to assess how much of a text is “human.”