It does not count as admission
You shouldn't flag it on that basis.
The reason is simple. The phrase "with the help of" and variations thereof is not a direct translation, nor logically related to, "I generated the following with". In other words, lacking further visible evidence in the post itself or in the OP's answer history of LLM usage, there is no proof that any of the subsequent code has been directly copy-pasted, or inconsequentially modified from an LLM. It's unreasonable for a moderator to delete a post solely based on that, and it's not warranted to flag it solely based on that.
If you do have additional evidence that a post was generated with an LLM beyond the alleged self-admission, please state that in your flag text. If you don't, thus suggesting that the alleged self-admission was the sole reason for flagging, and moderators don't find any other evidence by themselves, your flag is likely to be declined.
This is like someone saying "I found a solution thanks to this book [code block follows]" and someone else flags it as plagiarism. If the code in the answer isn't identical (or sufficiently similar to be considered unoriginal) to that found in the book, it's not possible to use plagiarism to justify its deletion.
A book, or an AI, could be instrumental in finding a solution without anything being copied out of it — the book, or the AI, was used as a guide, as inspiration, as a starting point; and the poster eventually came to an original solution on their own. In the spirit of assuming good faith it's not reasonable to construe a remark like "with the help of" as "the following is generated by", nor it's reasonable to flag based on that.
In fact, such a remark may very well be fluff. Quoting Thom A:
If your content isn't generated by ChatGPT, then any mention of ChatGPT is noise and doesn't belong in the post
If the post presents other issues (non-answer, link-only) it can be flagged and handled as such, as usual.
Variations
- with the help of [LLM]
- thanks to [LLM]
- I found the solution with [LLM]
- [LLM] pointed me to this
Unequivocally explicit admissions
Explicit admission of LLM usage is flaggable, but it must leave no reasonable doubt that the content of the post is LLM plagiarism. This leads to the next point:
The language problem
Yes, in the end it comes down to how you read the English language. Can we devise universally valid rules? I believe we can't. People with different English levels, different backgrounds, different native languages can and will interpret the same sentence differently. And this is why we can't assume that these remarks amount to admission. We don't know how the OP speaks English, and they might not mean "I copy-pasted the following from [LLM]". Deleting posts this way will lead to frustration and unfairness on all sides.
Ultimately, in these cases my advice is to focus on other indicators of AI usage. If these aren't available, flag conservatively. Eventually, ignore remarks of uncertain interpretation and edit them out.
If you are absolutely positive that an English sentence means the text was copy-pasted from an LLM, please do flag it, but in the end a moderator, who might or might not read English the same as you do, will have to decide what to do.
Keep in mind that the main goal, beside enforcing rules, is to do the good of the site.
The historical policy was given to moderators as a binding guideline for enforcing the ChatGPT ban. It doesn't strictly relate to community flaggers and curators. Moreover, even if it was repealed as a result of the community strike, explicit admission that some content is generated with LLMs is still actionable.
