Jump to content

Talk:Erythrocyte-based drug delivery

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

AI tag

[edit]

@MeowsyCat99: what makes you say this is AI generated? Czarking0 (talk) 13:50, 23 June 2025 (UTC)[reply]

Not necessarily entirely AI generated, but it is AI assisted. The first revision has very obvious signs of LLM assistance (title case in headings, the fact that it has two title headings, it uses a lot of the same wording/phrasing and tone ChatGPT uses). This article is large and I couldn't verify every citation (due to time) so I tagged it. MeowsyCat99 (meow) 14:04, 23 June 2025 (UTC)[reply]
"it uses a lot of the same wording/phrasing and tone ChatGPT uses" Can you give an example of this? I was under the impression it is very hard to tell these better than chance. To me, seeing that the images are marked as ChatGPT is the better tell? As far the the title stuff goes, I used title case in all the articles I wrote for the like first six months I was editing because I did not know the guideline. Czarking0 (talk) 23:06, 23 June 2025 (UTC)[reply]
Hi, yes. The images are a better tell. I didn't actually notice at first which was definitely my bad. None of these signs (e.g. title case or specific words) are guarantees that a LLM was used, and they shouldn't be used as some definite guarantee (which is why the template says may). They just help confirm suspicions as they pile up. It is mostly intuition, but I can definitely list some sentences that made me suspicious.
These systems fit within the broader field of nano-medicine and drug delivery by leveraging erythrocytes' natural ability to evade the immune system and circulate in the bloodstream for extended periods.
This topic is crucial for developing more efficient, biocompatible drug delivery methods, with applications in treating cancer, infectious diseases, and autoimmune conditions, improving drug efficacy while reducing side effects.
Targeting strategies play a crucial role in enhancing the efficacy, safety, and specificity of EBDDSs.
LLMs (specifically ChatGPT) use words like "broad[er]", "wide[ly]", "crucial", "highlight", "potential", "essential", "notable", etc. which help make the subject seem notable and important. It is easy to tell by the tone and phrasing they use. These words are fine and used a lot, but the presumptuous way LLMs specifically use it can be quite obvious.
Something else ChatGPT does is use "not only [...], but [also]", "is more than [just] a" or some variation of it, which happens in These systems not only enhance tumor targeting, but this approach leverages the inherent properties of erythrocyte membranes, [...].
But the thing that made me tag the article was:
This section will explore these drug loading methods in detail, highlighting their principles, efficience, and potential clinical applications.
where the LLM (I presume) refers to the text it is generating. MeowsyCat99 (meow) 07:28, 24 June 2025 (UTC)[reply]
Thanks, this is obviously an emerging area of editting. I wanted to dig into this more because I did the WP:NPP for this page and did not see these issues. To be honest, I still don't really see the broader, crucial issue. Do you have a source for that or is that your original observation? I somewhat agree This section will explore these drug loading methods in detail, highlighting their principles, efficience, and potential clinical applications. sounds a little LLMy but also a human may have wrote that. As you say the tag is uncertain; however, I would not be in favor of tagging every page that uses the phrases you underlined. So the question for me (and probably all of us collectively) is where to draw the line, even for the tag Czarking0 (talk) 13:56, 24 June 2025 (UTC)[reply]
Yes, neither am I in favor of tagging pages just because they use certain phrases or not. I tagged this based on intuition I developed from LLM texts I have dug through previously (having reviewed LLM-assisted school papers and read LLM generated online posts/articles). This is not some exact or formalized method. I don't have a source, it is anecdotal observation. I wish I could better formalize to you why this gives off very obvious LLM signs to me. You may find this resource useful. MeowsyCat99 (meow) 16:00, 24 June 2025 (UTC)[reply]

Move to draft

[edit]

@Onel5969: I see you moved this back to draft. Was there some discussion or consensus to do that? I searched WP:AfD and did not see it Czarking0 (talk) 18:32, 2 August 2025 (UTC)[reply]

When an article exhibits AI tendencies, if it is not improved, the appropriate course of action is to draftify, if it is potentially notable, which is the case here. Normally, I wouldn't move am article back to draft which has been promoted through AfC, but the tag was put on after the move to mainspace, and no significant work was done to address the issue. Onel5969 TT me 19:05, 2 August 2025 (UTC)[reply]
Ok moved to mainspace, the AI images were removed. Feel free to take it to or get consensus to move to draft. I am unaware of a policy on this but I don't think it is appropriate to unilaterally move things to draft based on on tags. Czarking0 (talk) 19:40, 2 August 2025 (UTC)[reply]