@drrachelbarr: AI detection tools are limited by the very mechanism that defines LLM text. They do not belong in academia. #neuroscience #science #chatgpt #student
The em dash is not on the keyboard. Period. Unless you know enough to use alt codes, no human can access the em dash on a keyboard. That red flag of seeing an em dash in a report, and saying ai wrote it is extremely valid.
2025-07-09 18:24:48
0
Lil préssec :
as a catalan native speaker doing a PhD, some of my papers got rejected by the plagiarism detector in part because of the fact that we are not english native speakers and sometimes we don't have that much of lexical or syntactic richness in front of a native speaker
2025-07-03 15:35:11
368
Cassiopeia(Autistic Version)🍉 :
As an autistic, I feel that "AI text" is sometimes really similar to my style. Not saying all autistic have the same flaws, but what feels like a "natural flow" to me often looks deconstructed (and somewhat underdeveloped or too much) by allistic folks... It shows how much we still don't know how to truly differentiate between ai vs human (especially due to diversity within humans)
2025-07-03 15:33:43
157
Doug Funny :
Teachers should just switch to the Socratic method and/or have students orally explain their research/papers and how they reached their conclusions
2025-07-03 16:33:43
2055
Melissa D. Gumbs :
I wrote a 750 word short story "from scratch" and had my Cabinet staff run it through an AI detector. It said it was over 50% riddled with AI. 😅 I'm sorry my vocabulary is expansive?!?
2025-07-04 14:39:37
162
Charles :
I would think highly stylized writers would also be at risk
2025-07-03 16:15:44
748
floralloveaffair :
I wrote a 10 page research paper before AI was really around. Out of curiosity I put it through AI detection a couple years later and it told me the paper was written by AI. This makes me nervous as a current adult college student!
2025-07-04 16:15:11
5
fredrik :
Ai programs archive 700+ sat score in maths. that's very high, and to say ai isbad education is missing the point.
what ai reveals to us that education of copy pasting data is useless as ai does it better. what education should be about is philosophy and critical thinking to use this tool to its fullest potential to combat false positives.
2025-07-23 08:10:22
0
custardandclove :
There are now AI tools specifically for making your text pass AI detectors. So Ai detectors are ironically promoting increased Ai usage by pushing honest students to use Ai in order to avoid wrongfully being accused of using Ai.
2025-07-04 00:10:35
37
Yuri Ontibon :
I completely agree, one false positive is too many. Thank you for sharing your knowledge.
2025-07-03 15:14:10
2430
Sarcastic_therapsid :
I hate that the “big indicator” is just when students write coherently.
2025-07-05 18:53:15
97
CJ :
It’s a tool. Just like a calculator. We need better ways to test intelligence than the regurgitation of information. Rather than banning a tool.
2025-07-20 21:57:09
3
Belinda Smith :
Yes, that’s because it’s normal, everyone uses it, you can’t use AI to determine it was AI or Not. ( ppl said the same thing for Computers and Calculators) It’s not the end of the world.
2025-07-04 07:11:30
4
Lee Forsythe :
I think the real problem here is that education relies too much on coursework. I don’t see any problem with students using ai to help structure or explain but if they rely too much on it, their degree will be pointless as they won’t have retained anything. I think the future will/should be less and less coursework based.
2025-07-03 15:56:26
2
Geff Harker :
The irony of the institution using AI tools to catch students from using AI tools is very late stage capitalism.
2025-07-05 06:12:22
160
Eli Fowler :
As a writer, I have an irrational hatred for the Em Dash that predates LLM's.
2025-07-04 06:41:45
11
Nuts_cracker :
The problem isn’t to LLM or not to LLM. Humanities education has come full circle, setting form over function and content. We mock our “backward” ancestors, but of the seven educational fields of the trivium+quadrivium, a typical student receives two and a half, with a gifted student selecting one partial additional subject.
2025-07-04 09:44:56
2
MP :
My writing is often flagged as AI.
2025-07-06 16:59:09
2
Crystal🫀 :
I was accused of using AI because my writing was “too professional”. I’m studying postgraduate medicine, I have 3 other degrees, and 2 decades of work experience. So… I should hope I’d be good at writing this stuff by now?! It’s very frustrating as someone who’s always put a lot of effort into producing high quality writing, and - hopefully! - ideas
2025-07-03 21:16:02
745
Larry.FF :
"em dash" is crucial to my life and no one can take it from me
2025-07-16 04:44:12
3
Susan :
Plagiarism is easy to detect. Force kids to write their essays in Word using tracked changes. From start to end. The final version should show an empty document being filled and every edit along the way. No AI can do this. And will never be able to do this.
2025-07-04 00:45:15
10
Luca Giovanni Solís :
You raise some valid points, but I disagree with the idea that AI detectors don’t work. They do work (though imperfectly) and many of them clearly warn about possible false positives, as you mentioned. Plagiarism is a serious issue, and dismissing or undermining detection tools doesn’t help address the broader academic challenges. Improving them, not discrediting them, should be the goal. Just to clarify, I'm just adding to the discussion here, not trying to argue, fight or anything like that.
2025-07-03 19:38:56
3
benji 🇵🇸 :
Literally one video above this one a college student was talking about how he got a 0% on a paper because he used an Em Dash.. his professor accused him of using AI even though he didn’t lol
2025-07-03 16:13:13
29
max_kain :
it's a lot like writing music. Software not even today, 25 years ago were pretty good at mapping an average techno song. But since most people don't really know the language they can't really detect it but I think they still can feel that something not genuine or like the lexic that doesn't correspond to the level of ideas. You can't have a basic techno groove and suddenly a chord progression that you would mostly find in Beethoven or classical/romantic era people are gonna notice but not if you are subtle tho. But overall you can find this discrepancy between the creative ability of a composer and the larger concept of his music and you will find no matter what that in 2-3 they're repeating a formula or if they're have a more developped language ( while still using formulas they are less obvious or feel less like formulas or have a greater range of lexic to express ideas in different ways) where as someone who gets they're software write their music for them will be uninspired and uninspiring after a few or even one song.
2025-07-03 20:54:21
2
To see more videos from user @drrachelbarr, please go to the Tikwm
homepage.