AI detection tools can be harmful

AI detection tools can be outright harmful when it comes to detecting the use of AI in writing. Here's why - and what I do instead.

Some are using AI detectors to spot AI-generated content. I've also heard of it happening in freelance writing circles, though I've never encountered it myself.

This UnfilteredFriday, let's talk about the problem of AI detector tools - and how they can actually do harm.

Defeating the checkers

Put yourself in the shoes of someone determined to use AI to produce work to pass off as their own. What would they do to avoid detection?

• Run it through another AI.
• Make manual changes.
• Introduce mistakes.
• Check against AI detectors.
• All of the above.

This means it's extremely hard for tools to detect AI use when users are actively trying not to be caught.

Highly unreliable

On the other hand, AI detectors are unreliable and could even unfairly penalise those who speak English as a second or third language.

This means that those who don't use AI have a high chance of being wrongly flagged, while those who use AI heavily might slip through unnoticed. And the innocent has a higher chance of getting wrongly accused.

Savvy humans are probably better at spotting AI use. As I've said before, heavy AI users are often the best at recognising AI-generated content, while those who don't use AI at all have no chance.

Should you AI?

As AI use grows, it's getting to the point where it's pointless to argue over whether a piece of content is AI-written.

Instead, here are some questions I ask when reading:

• Does it tell me something new?
• Did the author draw from their experiences?
• Is there obvious low-value AI dross?

These help me decide whether I’m reading something worth my time, or something I could’ve generated with AI myself.

And if I got value from it, then why should I care?

That said, I do get extra annoyed when reading low-value content that's AI-generated - I feel like I've wasted my time. And I use AI quite a lot.