Wikipedia:AI or not quiz - Wikipedia

From Wikipedia, the free encyclopedia

Test your AI detection skills! Determine whether each of the following Wikipedia articles are written by a human or an AI. Answers and signs of AI writing present are given at the bottom of each page. Note that some articles have been modified to remove obvious tells (like access dates) and the page history spoils the answer.

This test is not perfect: in the real world, the page history and the page creator's other edits should be considered.

Click on the links to access the texts.

/Scalping (trading)

OzzyMuffin238 9/10. The one about the rapa fruit dove was a real curveball. The common name of the animal wasn't emboldened, and the taxonomical name wasn't italicized. I was sure it was AI!…until I wasn't, that is.
Scopetpotffnaf2 7/10. Mostly pre-2020 human written articles, looked like AI at first sight. I thought I got the hang of it.
CrunchySpon 9/10. Only got the first one wrong, then I got the hang of it I think. Generally when the text is distributed in a bunch of small paragraphs instead of a few big ones it's AI generated.
Voussure 9/10. I mistook #4 as AI "writing," the bold text, writing style, bullet points, and large amount of quotations are what threw me off. But I didn't do as badly as I thought I would.
rgiyer32 7/10. All 3 I misidentified were pre-2020 human-written articles
KireiTonsi 7/10. I don't think I relied *too* heavily on guessing, but my confidence certainly wasn't consistently aligned with my accuracy whatsoever. Also, the answer drop-downs are coloured inconsistently which messed with me slightly.
Laxxius 5/10, I might as well have guessed, apparently I can't tell at all. I spent about 10 seconds on each one if that means anything
TheNerdy Parakeet 6/10, I might be too suspicious. Those AI ones are a little too good, though, we can write better that ChatGPT. Right, guys?
abigail 10/10, unfair advantage of being a comms grad as well as having to sit through conference speeches and slides that were beyond obviously AI
cup2/10 im so cooked
Hamimh2 (talk) 7/10! I thought some articles that were human were AI!! Humans, write better! Still great score though!
Luckytoday: 8/10. I think I was just lucky. — Preceding unsigned comment added by ~2026-24916-43 (talk)
Sddarealone (talk) : 7/10. I had few of my reasons correct for which I can't give spoilers here. The human ones quite honestly were surprising as I might have confirmation bias once few articles looked to be AI generated.
Mcrst.sean^(talk): 6/10. I thought 3 of the article we're AI, and one I thought was human.
Pwoli: 8/10. Both mistakes were AI false positives.
Philimania (talk) 12:06, 11 April 2026 (UTC): 7/10[reply]
User:HarrysAviation431 : 4/10
User:Randomperson112: 6/10; 2 of my mistakes were false AI positives, while the other two I mistaked for real human writing.
User:Edwardcap007: 9/10; had great difficulty checking signs, mostly focused on the references in the articles as well as wikitext formatting...
User:ElanorellëDuhíriel: 6/10; all four misses were false positives for AI writing. I might be too overly suspicious...
brachy08 _{(chat here lol)}: 6/10
HurricaneZeta^C: 8/10
Gnomingstuff: 8/8 (I contributed two examples)
Mox Eden: 5/10...
Ceayo: 8/10
User:Cooldood5555: 8/10
toby ^{(t)(c)(rw)(omo)}: 8/10 (I have a feeling us 8/10s lost to the same two examples lol)
Nil🥝: 8/10
Zzz plant: 7/10
Helpful Raccoon: 10/10
WeirdNAnnoyed: 8/10
Amendozu: 5/10
ChrysGalley 9/10 - though helped by personal knowledge on 3 items. And unhelped by personal knowledge of the item I got wrong....
🇵🇸‍🇺🇦 Fiddle^Timtrent Faddle^Talk to me 🇺🇦‍🇵🇸 7/10 with no tools such as GTP zero used
User:CR055H41RZ = 3/10,3 were correct as human/ai, 2 I focused on tiny details and mistook human for AI, 2 AI I missed, and 3 I got wrong. I felt like I was getting more used to the right and wrong answers by the end. (Bold letters)
the 🥭 man (the 🥭 talk and the the 🥭 contributions) 4/10! yikes...
A diehard editor (talk | edits) - 8/10
CabinetCavers:7/10 (3 human edits mistaken for AI)
jiraijohnny˚₊‧꒰ა ♡ ໒꒱ ‧₊˚ (KISS ME GOOD-BYE.⋆˚꩜｡): 7/10
User:starr_of_darkness: 8/10 I ended up doing better than i thought!!🤫🧏‍♂️
JustARandomSquid (talk) 09:22, 7 February 2026 (UTC): 7/10[reply]
Loki (talk) 07:22, 8 February 2026 (UTC): 6/10[reply]
PacificDepths 6/10
I don't know a good name.: 8/10 (failed on #8 and #10, I almost had it)
I am Transat: 9/10 (failed on the first one, got the rest correct)
Kadermonkey (She/Her): 8/10
🌀Hurricane Wind and Fire (talk) (contribs)🔥 15:34, 22 February 2026 (UTC): 8/10[reply]
User:Bluethricecreamman (Talk·Contribs) 23:21, 22 February 2026 (UTC): 6/10[reply]
✶Quxyz✶ 8/10, misidentifying Jewish peoplehood and Rapa fruit dove as AI. I think that those were the hardest, though.
Altoids0 (talk) 18:06, 4 March 2026 (UTC) 8/10, although that implies that I'm allowed to give myself the point for discovering Jewish Peoplehood in the wild [reply]
saraheileen76 6/10
—pythoncoder (talk | contribs): 10/10
MCPro1239953: 7/10
Smallangryplanet: 8/10
foxtail286: 7/10 (Yikes, these were tough!!)
Akbermamps: 7/10, oof. Felt like the last one could've gone either way.
Aperture541: 9/10. False-positive on "Jewish peoplehood". In hindsight, it was obvious, but I was a bit hung-up on the bullet points and the bolding.
Elkman ^(Elkspeak) 20:42, 16 March 2026 (UTC): 6/10.[reply]
Broc (talk) 15:57, 17 March 2026 (UTC) 9/10, it was fun[reply]
ClaudineChionh (she/her · talk · email · global) 00:48, 20 March 2026 (UTC): 7/10. Two false positives were the same as some of the spoilered ones above, though I did note one as "sloppy writing" so maybe I should rate this higher as an indication of humanity. The one false negative was one I noted as "Human spammer" – I think because at AfC or NPP I tend to decline or CSD for spam/promo reasons unless the AI is bleedingly obvious.[reply]
User:Kvng: 5/10. My detection is as good as a coin flip. 01:55, 20 March 2026 (UTC)[reply]
User:Phlogiston Enthusiast: 10/10. I have a sense for these things, I guess.
User:Wh3reitwill: 9/10; i failed on jewish peoplehood.
User:Gugalcrom123: 8/10; the two mistakes were falsely labelled LLM.
N51 DELTA TALK - 7/10, mistaken 2 pages for AI when they were human made, and 1 was mistaken for human made when it was AI made.
Fallbackintoreality (t • c): 5.5/10. I originally thought one of them was mixed, but it turned out that it wasn't. Two of them were real curve balls, but I also have to remember that this is the stuff AI is trained on.
✓ortexPhantom (talk): 8/10
Dafootballguy 8/10. I presumed a lot more AI then was actually there.
JeffSpaceman 4/10. I only got 1, 5, 9 and 10 right, which goes to show how much AI I assumed went into these pages. Perhaps I ought to brush up on my AI detection skills...
OsFish 9/10 I tripped up on Prevagen. But generally, I'm impressed by the Signs of AI page that helped me get 90% right.
Quofles 7/10 did better than I expected, but a lot of false positives on my part. Relied a bit too heavily on punctuation, bullet point markers. I need to focus a bit more on generalisations and lack of references in statements.
AmaSofiSofSof 8/10 honestly I think I was lucky to score so high, I have only skimmed the AI writing signs article, and I wanted to test myself without much prior knowledge on AI writing, I feel like a larger sample ( at least twice or 3 times as large should do) would give a much clearer picture.
Ace.Still.Life 8/10 main one that tripped me up was #8 because it looked so AI lolz, that was fun
Azurean72: 10/10! Quite surprised I managed to do so well, I was pretty split on a few of them.
Lea-Kim: 8/10 It was not as easy as I thought it would be. I tried to focus on the Avoidance of basic copulatives and it helped a lot.
8/9. I had accidentally viewed Aperture541's score about a false positive on Jewish peoplehood, so I had to skip that one. ozmoozmo@enwiki (talk:contribs) 08:23, 15 April 2026 (UTC)[reply]
User:Wooahitzy: 9/10, i was surprised i did so well !!
User:MangoMan11: 3/10... The human stuff looked AI generated to me and the AI stuff looked human. I do not think that many of these tells (i.e. red links and using other Wikipedia pages as a source) are very good for distinguishing AI from humans because I have encountered human editors making many of these same mistakes before
5/10. Humans, please write better! EinesteinNobel — $E=mc^{2}$ 22:41, 30 April 2026 (UTC)[reply]
RAK1999 : 7/10
User:EditWarCriminal 8/10, like many other users here I falsely marked the Jewish Peoplehood and Rapa Fruit Dove articles as AI-generated. I really should have remembered that people were in fact able to write lists with bolded words before 2022. As for the Rapa Fruit Dove one I think I just got suspicious of the unusually short yet verbose paragraphs, but I failed to consider it might just be a very unusual person!
Hey, Campssitie here, Got a 9/10. Not perfect, but not bad either! I tripped on The Jewish Peoplehood article being LLM-generated. Overrall, this was a blast to do! Thanks for all the wonderful people for making this test, and the people who did it, too. ★ Campssitie (msg) (contribs) 🧋🏖 19:31, 2 May 2026 (UTC)[reply]
9/10. I wish there was a larger corpus for this sort of thing, I don't like second-guessing myself. Thanks! Seercat3160 (talk) 01:15, 4 May 2026 (UTC)[reply]
Athanelar 8/10 for me. Finally got around to doing this; Rainforests of Malaysia got me because I was only looking for textual signs, I didn't realise this was also testing source integrity etc. I got a false positive on the Rapa Fruit Dove because of the "additional research is needed" bit in the lead; but had I bothered to read further than that, the rest of the text is pretty evidently human; so let's call it 8.5? Unlike others here, Jewish peoplehood didn't catch me out because to me its flowery prose was evidently of the human variety rather than the AI variety; I'm not really sure why. 05:42, 5 May 2026 (UTC)
CAROL.IS.BBO: 7/10 (Failed 3, 4, and 8). This was an interesting little "game" to play on my lunch break, but it did make me realize how much AI-generated content has influenced how I communicate. I can't seem to put an email together anymore without conjunctions / conjunctive adverbs (e.g. "but," "additionally") out the wazoo. I'm also putting together now why my most ChatGPT-reliant coworker uses so many em dashes in her texts..! (May 12, 2026)
BrackstheBrash: (2026-05-15) 8/10, failed 2 and 4... in part because I didn't finish reading the "Signs of" page until after I'd answered number 5.
Mr. Lechkar: 9/10 (failed #4). 27-05-2026
ArisNeta May 28th 2026: 9/10, the second one was what got me, admittedly after i failed that one i finally got my mind out of the gutter and understood how they were willing to play with me ... very fun!
7/8 29th may 2026 i panicked on the first question and guessed the wrong thing lol and then the jewish page got blocked by my college and i got spoiled for senchal, absolutely random college student without an account here
Obsidian the Fox: Spoiled myself for the first two (by looking at the history and at the source code, respectively), then got 6/8 on the remainder by misidentifying two of the human articles as AI, though it looks like the ones I got wrong were the two hardest ones. 23:26, 31 May 2026 (UTC)[reply]
Biangfox 9/10. The Jewishone caught me off guard
Theeddiebear888: 7/10 (2, 7, 8 wrong). Was suspicious of the short sections in #2, but ended up putting it down to bad writing. Thought that the shortness of #7 could be chalked up to the AI taking points directly from the reference, and #8 was tough! The header capitalization and poor writing made me want to say AI. I feel like I kinda understand what content to expect with AI, but I also might be tending more to the side of giving the benefit of the doubt when it comes to these articles. 20:25, 2 June 2026 (UTC)
Mac9kenny (talk) 15:35, 5 June 2026 (UTC)6/10 scary[reply]
UnitedKingdoms: 8/10. I got 4 and 8 wrong. It's helpful to read the WP:AISIGNS article as you play, and remember that bad or biased writing doesn't mean it was an AI. 21:05, 7 June 2026 (UTC)[reply]
Cedarshadow: 8/10. I recognized #4 from my time browsing articles covered in other Wikipedias. Passionate human meets low-traffic article. You got me on #6.

GPTZero: 10/10
Quillbot: 5 correct, 3 uncertain, 2 wrong
ZeroGPT: 7/10 (2 false positives, 1 "mixed" false positive)
Winston AI: 10/10

GPT-5.4: 8/10 (Prompted by ClaudineChionh on 2026-04-12.)
Opus 4.6: 10/10 (Prompted by ClaudineChionh on 2026-04-12.)

Can You Pass the Turing Test? (includes a version specifically about Wikipedia vs Wikipedia-style text)
Quiz by Winston AI (includes images and text)