Talk:Ferrock

Concerning my use of LLM assistance

1. My initial draft was human-written.
2. I asked ChatGPT to check for possible plagiarism, to clean up my grammar, and to remove statements that could be considered promotional (since Ferrock is a proprietary material).
3. ChatGPT detected several incidents of what it considered to be suspicious close paraphrase and awkwardness of expression.
4. I asked ChatGPT to remove the close paraphrase and to correct my grammar.
5. ChatGPT committed some horrible hallucinations in its suggested cleanup. For instance, a supposedly cleaned-up paragraph stated "The production process involves mixing these ingredients with water and exposing the mixture to carbon dioxide. This triggers a chemical reaction that forms iron carbonate, which hardens into a solid material." This is, of course, absolute total nonsense, mixing up details of the production process (grinding and mixing together the dry ingredients) with the application and curing (mixing with water and exposure to CO₂.
6. ChatGPT also flagged expressions such as "carbon-negative material" or "compressive strength ranging from 34.5 MPa to 48 MPa" as having been lifted almost word-for-word from the source material. Instead, I would state that these are standard phrasings in the engineering literature. My error (if there was one) was in not providing exact citations (providing only general ones), but my understanding is that one should provide exact citation if a statement is in any way controversial or otherwise can be challenged. "Carbon-negative material" appeared in multiple sources, so it is neither a unique phraseology nor is it in any way controversial. In the case of the range in compressive strength, the stated numbers appear in multiple sources which all seem to be citing some original source (which is probably one which I see is behind a paywall). Do you need me to dig up the original source? It would involve me going to the university library.

Prokaryotic Caspase Homolog (talk) 01:44, 3 November 2025 (UTC)