I recently raised a flag on a Question that's generated by an LLM (quoted at the end). The signs are clear, including the fact that the "Copy" sections remained, there are sections for the user to fill in (that is not code, so therefore not a duplicate of Should we flag human-written questions that use code generated by ChatGPT?), and the section of python code that "tried" was entirely missing their actual attempt. It was clear that it was generated by an LLM.

The flag, however, was declined as it being "about" an LLM. There is, however, no indication in the text it is about content from an LLM (such as "CrapGPT suggested this answer").

I flagged this again, because users aren't allowed to respond to flags any other way, further advising my conviction, and that a decline for such a flag raised in good faith isn't ok. It was declined again stating there is "no evidence" it is written by an LLM; the entire post is evidence enough.

This is a clear stance change from the moderator team, as I flagged a similar style post in the SG and that was deleted on the 12th. I can find no evidence of a change of stance posted here on Meta of the mods changing their stance.

Therefore I ask:

1. When did Questions entirely generated by AI become permissable, and why was this not communicated?
2. Why are Questions, like the below, flagged as AI generated declined, rather than disputed, when raised in good faith? This deters users from raising said flags, which only further encourages LLM content on the site, no?

Content of flagged post:

This is the exact content of the post, no adjustments have been made other than that made by putting the content in a quote block.

"I'm working on a project where I need to fetch data from severalrelated tables in a PostgreSQL database. The tables include millionsof rows, and my current query is slow. Here's a simplified version ofthe query:

sql Copy code SELECTa.name, b.detail, c.status FROMtable_a a JOINtable_b b ON a.id = b.a_id JOINtable_c c ON b.c_id = c.id WHEREa.created_at >= '2024-01-01'AND c.status = 'active'; I've already added indexes on a.created_at, b.a_id, and c.id. However, the query still takes over 30seconds to execute.

What I've tried:

Analyzing the query plan using EXPLAIN ANALYZE (but I'm not sure howto interpret some parts of it). Adding more specific WHERE conditions(minimal improvement). Experimenting with different types of JOINs.

I recently encountered the phrase 'What were you expecting?' in thecontext of [describe situation, language, or framework, e.g., Python,JavaScript, etc.]. Here's the scenario:

python Copy code

Example code causing the issue

result = some_function(input_value) print(result)

Error or unexpected output

The output I got was [insert output or error message]. Based on the documentation or my understanding, I wasexpecting [describe what you thought should happen].

sqlmysqlsql-server

Contents of flags

My text in italics, moderator in quote blocks.

This is clearly generated by an LLM, the "sql Copy code" and "python Copy code" and the sections for the OP to put their error cement it's pre-generated.

Declined - Questions *about* AI-generated content should be moderated or edited as usual instead of flagged. Please see: https://meta.stackoverflow.com/a/422440

I disagree this is about an LLM. There is place holder text here for the author to fill in and hasn't and no where do they mention this is from an LLM. At best my flag should be disputed, not declined, but I still disagree with this. If this is declined, I'll make a meta post about it, because a decline is wrong here.

Declined - There isn’t enough here to conclude that the whole question is AI-gen. The only markers are around code blocks, it’s reasonable to assume it’s a question “about” AI code