25
$\begingroup$

This answer claims that eigenvalues of a matrix are preserved under elementary row operations.

It has two upvotes and it is an accepted answer. I came across it while searching related topics on Google. I am worried this answer might mislead learners like me.

The user does not seem active so I did not comment on the answer.

I wanted to bring it to attention because the language of the post also seemed AI-generated... "Subtle non-sense is the worst kind of non-sense"...

$\endgroup$
14
  • 4
    $\begingroup$ Amusingly, GPTZero reports a likelihood of 88% for AI generation (not that it's foolproof, of course, but I have seen very similar styles of answer from GPT in the past, including the bulleting and the inconsistent use of LaTeX). And yeah, the eigenstuff fails under trivial examples, and several others are trivial or just not noteworthy in the least IMO. [cont.] $\endgroup$ Dec 28, 2023 at 19:55
  • $\begingroup$ Poring through their recent posts this one (claimed likelihood of 95%) was also called out for being GPT-generated; one wonders what one would find among their deleted posts. [cont.] $\endgroup$ Dec 28, 2023 at 19:55
  • 11
    $\begingroup$ IIRC, GPT is not allowed anymore, so one could flag posts of theirs that appear to be AI-generated. They do seem to be active ("seen in the last week"), as well. And honestly, comments correcting them at minimum are perfectly kosher, IMO -- even if the user won't respond to them or address them adequately, it can let onlookers know something's amiss (a services I've been thankful for several times in the past). $\endgroup$ Dec 28, 2023 at 19:55
  • 1
    $\begingroup$ @PrincessEev Oh I missed that... July 29 was the last time they posted something so I thought that they are inactive. $\endgroup$ Dec 28, 2023 at 20:01
  • 3
    $\begingroup$ @PrincessEev You are right... A comment will prevent others from being misled. I made a comment with explanation. $\endgroup$ Dec 28, 2023 at 20:18
  • 5
    $\begingroup$ Dear mods: 1) Can you confirm if Eev is right that GPT is no longer allowed? And 2), what's the procedure for flagging answer(s) as GPT? It's not offered as a flag reason. $\endgroup$
    – JonathanZ
    Dec 28, 2023 at 20:52
  • 1
    $\begingroup$ @JonathanZ - See math.meta.stackexchange.com/questions/35342/… or math.meta.stackexchange.com/questions/35881/… $\endgroup$
    – mr_e_man
    Dec 28, 2023 at 22:23
  • $\begingroup$ @mr_e_man - Aren't those both pre-strike? They're also kind of unclear, and I know I would appreciate a simple yes/no and "do this" answer. Maybe I need to post a separate question to ask about those, but this seemed like a good occasion to ask, as we're dealing with a specific actual instance, not hypotheticals. $\endgroup$
    – JonathanZ
    Dec 28, 2023 at 22:38
  • 2
    $\begingroup$ The answer referred to in the body of this question has been deleted by votes of three users. $\endgroup$ Dec 29, 2023 at 2:38
  • 1
    $\begingroup$ Somewhat related: Should we vote to delete wrong answers? Some of the linked questions might be related, too. $\endgroup$ Dec 29, 2023 at 10:22
  • 3
  • $\begingroup$ A.I will be the death of mathematics education .... $\endgroup$
    – K.defaoite
    Jan 7 at 17:27
  • 2
    $\begingroup$ Go to the physics site, and you can find upvoted wrong answers and accepted wrong answers until the cows come home. $\endgroup$ Jan 10 at 12:07
  • 1
    $\begingroup$ @SuzuHirose the thing about this wrong answer was that it was blatant false information and not some logical inconsistency. [I haven't studied physics since high school so can't comment on that :)] $\endgroup$ Jan 10 at 18:08

1 Answer 1

7
$\begingroup$

The answer referred to in my question has been deleted after downvotes.

enter image description here

In my opinion, users may use ChatGPT or other AI tools.

  • However, the specific question they are answering should be within their knowledge, expertise or learning.
  • They should make sure that they are posting correct answers.

For otherwise, blatantly wrong answers will mislead people who are trying to learn.

Incorrect AI-generated answers are often hard to spot because they are perfect blend of nice language and subtle nonsense. (Hallucination) This could be why that answer was accepted and it got two upvotes as well...

  • Answers with too many words should be critically reviewed based on substance of the information before upvoting.
$\endgroup$
1
  • 37
    $\begingroup$ My answers are a blend of nice language and subtle nonsense, and I don't even need advanced technology to write them. $\endgroup$
    – MJD
    Dec 30, 2023 at 0:20

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .