AI Policy

Although I sometimes use AI for idea generation and structural planning, as a rule, I don't use AI tools for writing. However, I will not work with clients who overly rely on AI-checking services. The reason? They don't work.

Regardless of what the service providers themselves say, AI-checking services are notoriously unreliable and many studies have found they overwhelmingly produce false positives. I've also run my own tests to confirm this. For example, I've had entirely human-written content flagged as 100% AI-generated, then changed two sentences (out of 7,000 words), and the same checker has revised that down to 10% AI-generated. Conversely, I've seen 100% AI-written content confidently judged to be 100% human-written.

I won't name and shame, but let's just say I've tested virtually every AI checker currently available. I also strongly suspect that at least some of these services are deliberately flagging content so that they can profit from so-called 'Humanizer' services - either because they are directly owned, or through affiliate deals. Anyway, that's something for my blog posts.

The point is that I will not invest hours or days writing a piece of content only to have it arbitrarily judged by completely flawed and unreliable technology.