Hacker News Reader

> Just a few weeks ago they released a tool to "detect" ChatGPT. Obsolete already?

I've seen so much hype around these tools. Not only are they theoretically unsound, they're downright dangerous and equip folks with spurious confidence. Going forward, the default assumption should be that the content you're looking at is fake unless you have sufficiently high trust in the source.

The only robust human content verification methods I’ve heard of are interrogating the content creator afterwards to see if they can adequately explain what they wrote.

My friends in law school are telling me there's been an emergency pivot away from "take home" exams back to "in class" exams.

Schools are obsolete if they want to use these tools.

The world has changed and their job is to prepare students for it.

I have no confidence they've achieved an acceptably low false positive rate.