Timeline for How many natural operations on subsets are there?
Current License: CC BY-SA 4.0
16 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Sep 28 at 18:57 | history | edited | Martin Brandenburg | CC BY-SA 4.0 |
added 300 characters in body
|
| Sep 27 at 14:54 | history | edited | LSpice | CC BY-SA 4.0 |
Adding which Peter, since there is now another in the thread
|
| Sep 27 at 14:31 | comment | added | Martin Brandenburg | Based on the criticism and the numbers of downvotes, I have removed the long list, the reference to ChatGPT, and instead used Peter's concise description. This makes some of the comments here obsolete, and I am sorry for that, but I think the answer is better now. | |
| Sep 27 at 14:30 | history | edited | Martin Brandenburg | CC BY-SA 4.0 |
removing ChatGPT's list, use Peter's description
|
| Sep 27 at 14:12 | comment | added | Peter Taylor | @user509184, until "probably-correct" can be replaced with "correct", the person using the tool should do the checking before posting it and asking every reader to do it. But once you account for generating and correcting it's quicker to write the Python script. This is a fundamental case of the wrong tool for any job where correctness matters. That's separate from the point that the prompt given to ChatGPT likely expresses the 16 cases in a more readable format than this output. | |
| Sep 27 at 13:54 | comment | added | LSpice | @MartinBrandenburg, re, sorry, you are right; I forgot this behavior of spoilers. I thought they were like folding blocks. | |
| Sep 27 at 13:34 | comment | added | Martin Brandenburg | I did not use a Spoiler tag since afaik they grow with the content? In any case, if someone finds a nice formatting, feel free to edit. | |
| Sep 27 at 13:25 | comment | added | Martin Brandenburg | Indeed I just wanted to quickly get an answer (from the classification of deflating maps which ChatGPT did flawlessly), and I am aware that I could also write a script, but I thought this is a good task for ChatGPT. I was wrong. Lesson learned. Thanks for fixing the duplicate! | |
| Sep 27 at 13:20 | comment | added | user509184 | @PeterTaylor To me this situation looks like the "best-case scenario" that people sometimes propose for the use of LLMs for mathematics: a small, systematic task, with finite output that can be quickly checked, such that one could give a LLM a natural-language prompt to generate probably-correct output in less time than one could write a little script, in Python or whatever, to generate the correct output. I think it is remarkable that even in this scenario, ChatGPT still stumbled, so that in the end it would have been faster to just write the script. A worthwhile experiment IMO. | |
| Sep 27 at 13:09 | comment | added | LSpice |
In terms of avoiding scrolling, you can always (ab)use spoilers, entered blockquote-like as >! (see MO syntax). \\ In addition to #12 and #16 pointed out by @PeterTaylor, #11 and #15 were the same. I edited to correct.
|
|
| Sep 27 at 13:08 | history | edited | LSpice | CC BY-SA 4.0 |
Fixing redundancy pointed out by @PeterTaylor (https://mathoverflow.net/posts/comments/1248406)
|
| Sep 27 at 12:04 | comment | added | Peter Taylor | 12 and 16 are the same. Why use ChatGPT for something like this when it's easier to write a program to generate it in any half-decent programming language? Or, indeed, it can be expressed more compactly without using any other tools: $\alpha(\emptyset, \emptyset) = \emptyset$; $\alpha(\emptyset, V)$ is either $\emptyset$ or $V$; $\alpha(U, \emptyset)$ is either $\emptyset$ or $U$; and $\alpha(U, V)$ is one of $\emptyset, U, V, U \cup V$ where $U$, $V$ are non-empty in the cases where mentioned and the cases are independent, giving $16$ options. | |
| Sep 27 at 11:43 | history | edited | LSpice | CC BY-SA 4.0 |
Numbering; `\template`
|
| Sep 27 at 11:35 | history | edited | LSpice | CC BY-SA 4.0 |
Numbering
|
| S Sep 27 at 11:31 | history | answered | Martin Brandenburg | CC BY-SA 4.0 | |
| S Sep 27 at 11:31 | history | made wiki | Post Made Community Wiki by Martin Brandenburg |