Show HN: GPTMinus1 – Fool OpenAIs AI Detector by Randomly Replacing Words https://ift.tt/6mVfKH9
Show HN: GPTMinus1 – Fool OpenAIs AI Detector by Randomly Replacing Words This is a joke before anyone gets too worked up. The algo is a simple one which goes and replaces random words with their synonyms. It will obviously beat these AI detectors but the output isn't probably what you would want. The point stands though that more sophisticated approaches that rephrase or mix up the word distribution will be able to throw off the detector and it should not be relied on exclusively for, say, checking if a student assignment is AI generated. https://ift.tt/2fc9pAM January 31, 2023 at 11:26PM
Labels: Hacker News
0 Comments:
Post a Comment
Subscribe to Post Comments [Atom]
<< Home