News for hackers. Stuff that matters.
Re:AI Detectors Falsely Accuse Students of Cheating–With Big... by torginus on Monday October 21, 2024 @05:55AM (#41902432)
Just because a method of assessment became easily spoofable doesn't mean we should give up on it. Imagine if in the era before HTTPS we just said that the internet won't be really viable because it's impossible to communicate securely on it.

I still feel like AI detectors would work well if we have access to the exact model, output probabilities of tokens, We can just take a bit of given text, and calculate the cumulative probability that the AI would complete it exactly like that.

Re:AI Detectors Falsely Accuse Students of Cheating–With Big... by nick3443 on Monday October 21, 2024 @06:04AM (#41902486)
Probability is not an acceptable way to determine a student's future. They may have learned from the AI and remember some of the exact phrasing, and learned writing/language cues from it as well.
Re:AI Detectors Falsely Accuse Students of Cheating–With Big... by alias_neo on Monday October 21, 2024 @09:01AM (#41903714)
Agreed. I'm not a good writer, tending to stick to a somewhat abrupt, point-making structure almost better suited for bullet pointing. I've taken tips from other HN users on how to improve, but I have no doubt that had I been going through university these days, I'd probably be flagged too.