Hacker News Reader

Comment by hilux | Hacker News Reader

These AI detectors have all the validity of phrenology. You can enter some of your own writing (I did) to easily see for yourself. The text I entered had been online, and still is online, longer than AI writers have been around - but that doesn't seem to feed into their logic.

View on HN

A good example: Put bits of the US Constitution in some AI detectors and they'll tell you it was AI generated. https://arstechnica.com/information-technology/2023/07/why-a...

Interesting--perhaps the AI detectors are right, and this will be how we discover that we are in fact in a simulated universe!

...or at least a simulated democracy.

The whole idea of an "AI detector" is stupid. If it ever does work, it will only work until someone trains a model against it.

And anyway, text exists to be read by humans. If the buyer needs an “AI detector” to tell if the copy was AI generated, the AI would seem to be good enough.

I think this only exists to facilitate some already-bad heuristics, like the idea that nobody in-house needs to check the contractor’s work-product.

Which text did you enter, and into which AI detector?

I tried it with my blog, and it was correctly identified by Copy Leaks as written by a human. I don't know how representative https://copyleaks.com/blog/ai-detector-continues-top-accurac... is, but certainly all of them are getting above phrenology detection rates.

Even with this guy, the problem wasn't that it misidentified him as AI. It was that "we can't tell" got turned into, "we can't risk Google thinking he might be AI".

""we can't tell" got turned into, "we can't risk"

Oh god, we've developed yet another shitty lie detector.

You only just noticed that?

Reply on news.ycombinator.com