Humanities desk
< March 28	<< Feb \| March \| Apr >>	March 30 >

Welcome to the Wikipedia Humanities Reference Desk Archives
The page you are currently viewing is a transcluded archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

March 29

Looking for a paper that I think ChatGPT has invented

I was experimenting with ChatGPT for the first time and asked it to research a particular aspect of finance (whether there is a link between how often investors check their portfolios and their performance). ChatGPT gave a very convincing response including several citations to academic papers. The trouble is that I think it made them all up. One example was "Christophe Faugère and Julien Pénasse, (2017). The more we know, the less we think we know: The role of information acquisition in decision making. The Journal of Behavioral Finance, 18(1), 1-11. https://doi.org/10.1080/15427560.2017.1273352". Faugere and Penasse exist but the link is dead, and that particular issue of the Journal of Behavioral Finance appears to have a completely different article on the cited pages. Googling showed no sign of this paper either. I'm aware that ChatGPT often makes stuff up but I didn't know it invented scholarly papers complete with fake citations. Is it possible that this article does exist and I'm just looking in the wrong place? Thanks for your help. --195.206.172.158 (talk) 09:16, 29 March 2023 (UTC)[reply]

ChatGPT certainly does make up citations. Here is just one of the hits I get from searching for "ChatGPT inventing citations". The bot simply learns what things have gone together: it has no knowledge or understanding of the scope of an item that needs to be kept together (i.e. that a whole citation must be treated as an unalterable object). ColinFine (talk) 09:49, 29 March 2023 (UTC)[reply]

See also here, here, here, etc. Shells-shells (talk) 09:54, 29 March 2023 (UTC)[reply]

Thanks for doing my Googling for me guys :-). (Particularly enjoyed the made-up study from Kim, Kim, Kim, Kim, Sausage, Eggs and Kim. It reminded me of the Welsh rugby team some years back which consisted mainly of Jones lining up alongside Jones and Jones with Jones, Jones and Jones at the base of the scrum.) I was aware it made stuff up but this seemed so specific, and it's been so long since I looked up an academic journal that I didn't trust myself.

There was one article it cited that was entirely real, but didn't seem to support its conclusion. I asked it to provide more detail of where the article contained the information; it gave quotations in inverted commas and page numbers, all made up. Conclusion: never trust anything ChatGPT says that involves academic papers. --195.206.172.158 (talk) 10:02, 29 March 2023 (UTC)[reply]

It's important to remember that ChatGPT and similar programs are all natural language simulators, designed to create cromulent English passages of text motivated by the input of the user. What they don't do is understand the text they are creating, it's all basically "Colorless green ideas sleep furiously..." type writing, but with the power of a sophisticated AI. Which means that while it can create plausible fiction, what it cannot do is compile research properly or interpret and rephrase technical information. This becomes plainly obvious as any of the technical writing it does may look reasonable to laypeople, but in my experience it always reads like "colorless green ideas sleep furiously" to anyone with a modicum of familiarity with the subject. --Jayron 32 15:24, 29 March 2023 (UTC)[reply]

Kim, Kim, Kim, Kim, Sausage, Eggs and Kim" reminds me of the Spam skit from Monty Python. --User:Khajidha (talk) (contributions) 02:16, 30 March 2023 (UTC)[reply]

One shudders to think of the possible results of training ChatGPT on the Monty Python corpus. ("This corpus is dead"?) -- Verbarson ^talk_edits 19:16, 31 March 2023 (UTC)[reply]

It's an ex-corpus. Mathglot (talk) 07:14, 5 April 2023 (UTC)[reply]

I do not like green Eggs and Kim! —Tamfang (talk) 03:36, 5 April 2023 (UTC)[reply]

Chapter and Verse designation for Books of the Bible (OT & NT)

For books like Exodus and Mark (to pick a OT & NT book that aren't apocrypha to anyone), when was the current chapter and verse used in Catholic/Protest/Greek Orthodox, etc agreed to? Was the chapter and verse of Exodus established before the meridian of time by the Jews and picked up by the Christians? Are there any Christian Groups that use a different Chapter and/or verse structure?) I know that Psalm 151 used by the Greek Orthodox is sort of an exception. I'm basically wondering whether there is any Christian group where Genesis 19:12 isn't the angels talking to Lot or John 3:16 isn't "For God so loved the world..."

(This is coming from a discussion of trying to create an equivalent to the Bibleverse Template for the Book of Mormon and the fact that a chapter/verse structure wasn't made by the CoJCoLDS until well after Brigham Young took them to Utah) Naraht (talk) 14:41, 29 March 2023 (UTC)[reply]

WHAAOE. See Chapters and verses of the Bible, which discusses the development of the chapters and verses for the Bible. The Old Testament versification was largely fixed in its modern form by about the mid 15th century, with the current system of versification of the New Testament reaching its fixed form about a century later. --Jayron 32 14:51, 29 March 2023 (UTC)[reply]

The numbering of the Psalms does vary rather. DuncanHill (talk) 14:56, 29 March 2023 (UTC)[reply]

There are also other idiosyncrasies. The ending of the Gospel of Mark, for example, Mark 16, has at least three different endings: The "no ending" version that ends abruptly at Verse 8, the "shorter ending" which extends verse 8 (and which some consider an "unnumbered verse") and the "longer ending", which includes the verses commonly numbered 9-20 (and for which some versions also append the so-called "shorter ending" after verse 20). Most bibles I have seen do include all of this text, but it is often notated as being not universally accepted as canonical. There are also smaller controversial sections, including the Johannine Comma (1 John 5:7b-8a), the Pericope Adulterae (John 7:53–8:11) and several others which are often numbered along with the rest of the text, but omitted from several bibles; this can lead to a breaking of the standard verse numbering; bibles that omit these passages from the main text don't renumber the verses, but usually have a footnote explaining the discrepancy. Others include all of the verses, but include footnotes explaining the controversies. Wikipedia has additional articles on these topics, such as Textual variants in the New Testament, Textual variants in the Hebrew Bible, List of New Testament verses not included in modern English translations, etc. --Jayron 32 15:12, 29 March 2023 (UTC)[reply]

The verse divisions of the Hebrew Bible are actually somewhat old (though not dating back to the Biblical period), since they were the foundation for the cantillation systems which were probably developed starting in the 700s A.D. (though the earliest surviving materials on such mainly date from the 900s A.D.). Everything else (verse numberings, chapter divisions and numberings, and New Testament verse divisions) apparently dates from well after 1000 A.D. AnonMoos (talk) 17:16, 29 March 2023 (UTC)[reply]

There's also older weekly Torah portions bigger than the chapters but only in Genesis through Deuteronomy. Sagittarian Milky Way (talk) 23:16, 30 March 2023 (UTC)[reply]

Historic photo

While this appears to be the original photo, with borders and creases, this one, despite being essentially the same, features more details and the bottom part that is absent on the first photo. How is that possible? Judging by identical postures, I doubt that two separate photos of the same group have been taken. And this is newspaper reproduction most likely. Brandmeister^talk 22:13, 29 March 2023 (UTC)[reply]

The images could be of two different prints of the same original negative. --174.89.12.187 (talk) 23:24, 29 March 2023 (UTC)[reply]

Or Photoshop (other image-editing software is available). Alansplodge (talk) 09:54, 30 March 2023 (UTC)[reply]

The photo with more contrast has a strip at the bottom that is cut off in the faded photo. The rightmost part of that strip shows a somewhat fuzzy horizontal band, as if someone has tried to erase a caption or watermark. Perhaps this image was distributed as a postcard. --Lambiam 11:26, 30 March 2023 (UTC)[reply]

Yes, it had a watermark that was erased. Brandmeister^talk 13:00, 30 March 2023 (UTC)[reply]