-4
$\begingroup$

After and during writing the question on MSE, some similar posts are suggested by the site. However, there are still many duplicate questions posted every day. I think it would be a good idea to provide a means to detect duplicate posts automatically and better. For example by automatically searching for similar posts with means like Approach zero or even by using artificial intelligence (e.g. LLMs) to search for similar posts based on text, formulas and images.

Or even one step ahead, the AI model suggests a post that is not a duplicate of the question, but the method for solving it is relevant. I think there is good potential to improve this aspect of the site.

$\endgroup$
4
  • 3
    $\begingroup$ Personally, I'd object to having AI systems scan for duplicates. Every question is a duplicate, really. Or most of them anyway. I doubt we could persuade the system to differentiate between the duplicates we tolerate and those we don't. $\endgroup$
    – lulu
    Mar 29 at 1:18
  • 12
    $\begingroup$ I am actually in agreement that the on-site search is terrible, and that approach0 is only a bit better. Better search would be great. I am not sure, however, that the hot new hotness of "generative AI" and "LLMs" is the answer. $\endgroup$
    – Xander Henderson Mod
    Mar 29 at 1:51
  • 4
    $\begingroup$ If this could be done, it would already be happening on Stack Overflow. It definitely won't happen here first, and it would be a network-wide development if it could happen at all. $\endgroup$
    – Nij
    Mar 29 at 7:28
  • 3
    $\begingroup$ There is no "artificial intelligence" in existence yet. You have stochastic parrots and generative neural nets. Any notion that there is "intelligence" there is pure hype. $\endgroup$ Mar 30 at 16:03

1 Answer 1

3
$\begingroup$

This would be great to have, but it's harder to do than it sounds. It's an issue that affects not just Math.SE, but many Stack Exchange sites, including Stack Overflow, and it's not straightforward to solve it. The challenge is finding duplicates is non-trivial, and LLMs and AI aren't a magic silver bullet. So I think the goal is great, but it's not clear how to achieve it right now.

$\endgroup$

You must log in to answer this question.

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