frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Students using “humanizer” programs to beat accusations of cheating with AI

https://www.nbcnews.com/tech/internet/college-students-ai-cheating-detectors-humanizers-rcna253878
43•unpredict•3d ago

Comments

ashleyn•1h ago
I'm guessing this "humanizer" actually does two things:

* grep to remove em dashes and emojis

* re-run through another llm with a prompt to remove excessive sycophantry and invalid url citations

dbg31415•1h ago
You’re absolutely right!

Ha. Every time an AI passionately agrees with me, after I’ve given it criticism, I’m always 10x more skeptical of the quality of the work.

glitchcrab•59m ago
Why? The AI is just regurgitating tokens (including the sycophancy). Don't anthropomorphise it.
20260126032624•53m ago
Because of the way regurgitation works. "You're absolutely right" primes the next tokens to treat whatever preceded that as gospel truth, leaving no room for critical approaches.
otikik•15m ago
Because I was only 55% sure my comment was correct and the AI made it sound like it was the revelation of the century
the_fall•1h ago
No. No one is looking for em-dashes, except for some bozos on the internet. The "default voice" of all mainstream LLMs can be easily detected by looking at the statistical distribution of word / token sequences. AI detector tools work and have very low false negatives. They have some small percentage of false positives because a small percentage of humans pick up the same writing habits, but that's not relevant here.

The "humanizer" filters will typically just use an LLM prompted to rewrite the text in another voice (which can be as simple as "you're a person in <profession X> from <region Y> who prefers to write tersely"), or specifically flag the problematic word sequences and ask an LLM to rephrase.

They most certainly don't improve the "correctness" and don't verify references, though.

emmp•1h ago
For student assignment cheating, only really the em dashes would still be in the output. But there are specific words and turns of phrases, specific constructions (e.g., 'it's not just x, but y'), and commonly used word choices. Really it's just a prim and proper corporate press release style voice -- this is not a usual university student's writing voice. I'm actually quite sure that you'd be able to easily pick out a first pass AI generated student assignment with em dashes removed from a set of legitimate assignments, especially if you are a native English speaker. You may not be able to systematically explain it, but your native speaker intuition can do it surprisingly well.

What AI detectors have largely done is try to formalize that intuition. They do work pretty well on simple adversaries (so basically, the most lazy student), but a more sophisticated user will do first, second, third passes to change the voice.

postepowanieadm•1h ago
The Washing-Machine Tragedy was a prophecy.
falloutx•1h ago
At some point, writing 2 sentences by hand will become more acceptable than this.
pinnochio•1h ago
Shortly after, AI-powered prosthetic hands that mimic your handwriting will write those 2 sentences for you.
yarrowy•1h ago
just move to 2 hour in class writing blocks.
tgrowazay•1h ago
Everyone knows about emdashes, but there are so much more!

Here is a wiki article with all common tell-tales of AI writing: https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing

jurgenaut23•1h ago
I used em dashes heavily 15 years ago when writing my PhD thesis.
SecretDreams•1h ago
Same, but 5 years ago. Now they're ruined for me lol.
singpolyma3•36m ago
So did every author of classic literature. People who think they can spot AI writing by simple stylistic indicators alone are fooling themselves and hurting real human authors
zeroonetwothree•21m ago
Let’s just say when my coworkers started sending emails with tons of bold and bullet points when they had never done that before I felt pretty justified in assuming they used AI
AstroBen•48m ago
I saw someone created a skill to weaponize that exact list to humanize the AI's output

There are no clear signs, at least for anyone who cares to hide them

OsamaJaber•1h ago
AI detectors punishing non native English speakers for writing too cleanly is the part nobody talks about enough -_-
Rexxar•12m ago
For example, native English speakers often make phonetic spelling errors (such as its/it’s, your/you’re) that non-native English speakers usually avoid. It’s probably a sign that someone speaks more fluently when he starts making these types of mistakes from time to time.
HeavyStorm•54m ago
This ship has sailed.

It's how it was with the internet. I grew up in the 90s, and teacher didn't know how to deal with the fact we no longer had to go through multiple books in the library to get the information they needed. We barely needed to write it.

Now nobody expects students to not use the internet. Same here: teachers must accept that AI can and will write papers and answer questions / do homework. How you test student must be reinvented.

xeromal•49m ago
I know a lot of teachers are reverting back to handwritten papers. People can generate it but at least you're doing something.
randall•44m ago
this is like irl cryptographic signatures for content lol
idiotsecant•42m ago
This is just about the worst possible response it seems. It manages to probably hurt some wrists not used to long handwriting sessions, completely avoid learning how to use and attribute AI responsibly, and still probably just results in kids handwriting AI generated slop, anyway.
singpolyma3•37m ago
While I don't think it's the right solution, it will force them to at least read what they're submitting which means some learning :)
BugsJustFindMe•36m ago
> It manages to probably hurt some wrists not used to long handwriting sessions

I'm sorry but, lmao. You cannot be serious.

> attribute AI

Oh no!

> still probably just results in kids handwriting AI generated slop

Not if they're doing it in person. And at least they then need to look at it.

jxf•25m ago
We've been writing with our hands for thousands of years. I suspect that on balance a Butlerian Jihad against AI slop would be perfectly fine for our hands.
ethin•9m ago
It also disadvantages people with disabilities. How exactly are they supposed to do these papers and tests? Dictate everything to someone else, using Blindness as an example? Because that seems very very inefficient and extremely error-prone.
smoyer•36m ago
When I was in high school, we were not allowed to use calculator for most science classes ... And certainly not for math class. I'm ten years, will you want to hire a student who is coming out of college without considerable experience and practice with AI?
croes•21m ago
Why would I want to hire such a student? What makes him better the better pick than all the other students using AI or all the other non-students using AI?
AlotOfReading•20m ago
LLMs work best when the user has considerable domain knowledge of their own that they can use to guide the LLM. I don't think it's impossible to develop that experience if you've only used LLMs, but it requires a very unusual level of personal discipline. I wouldn't bet on a random new grad having that. Whereas it's pretty easy to teach people to use LLMs.
nkrisc•17m ago
If all they know is AI, and they supplanted all their learning with AI, why even hire them? Just use the AI.
ThrowawayR2•11m ago
[delayed]
AndrewKemendo•33m ago
I remember when websites couldn’t be considered valid sources for graded assignments
MattGaiser•30m ago
I was dealing with this even in 2014 when I was in high school. Even then, entire classes of government data weren’t published in some print volume.
AndrewKemendo•26m ago
In my case at least there was some validity to it in 1995
croes•17m ago
This is not how AI will be in the future.

At some point the will have to make profit, that will shape AI.

Either by higher prices or ads. Both will change the use of AI

mc32•39m ago
I get that students are using the LLM crutch -and who wouldn’t?

What I don’t get is why wouldn’t they act like an editor and add their own voice to the writing. The heavy lifting was done now you just have to polish it by hand. Is that too hard to do?

BugsJustFindMe•31m ago
Humans tend to be both lazy and stupid and are always looking for ways to pass by with minimal effort. Kids aren't different just because they're in school.
MattGaiser•27m ago
It would be dull to do. Being a tone scribe would be terrible.
zkmon•37m ago
Teachers are also heavy users of AI. Entire academic business staff is using AI.

The goals of academic assessment need to change. What are they assessing and why? Knowledge retention skills? Knowledge correlations or knowledge expression skills? None of these going to be useful or required from humans. Just like the school kids are now allowed to use calculators in the exam halls.

The academic industry need to redefine their purpose. Identify the human abilities that are needed for the future that is filled with AI and devices. Teach that and assess that.

kyykky•22m ago
Teaching is about moving our knowledge (the stuff we’ve collectively learned ourselves, from others and our parents [instead of everyone needing to find out on their own]) to the next generation. While some skills may become obsolete in some parts of professional life due to AI, the purpose of academia does not change much.
croes•18m ago
Teachers are also heavy users of solution books, but would not give them to students for this reason.
Espressosaurus•7m ago
A calculator is more consistent and faster at calculating than I am, but I still need to understand how to multiply, divide, add, and subtract before I can move on to more complicated math. I need to intuitively understand when I'm getting a garbage result because I did an operation wrong, moved a decimal place by accident, or other problem.

Memorization has a place, and is a requirement for having a large enough knowledge base that you can start synthesizing from different sources and determining when one source is saying something that is contradicted by what should be common knowledge.

Unless your vision of the future is the humans in WALL-E sitting in chairs while watching screens without ever producing anything, you should care about education.

threemux•26m ago
In person, proctored blue book exams are back! Sharpen those pencils kids.

I've been wondering lately if one of the good things to come out of heavy LLM use will be a return to mostly in-person interactions once nothing that happens online is trustworthy anymore.

bambax•18m ago
Yes! This "problem" is really easy to fix with in person exams and no computers in class, ever.
idle_zealot•16m ago
But is that webscale?
meroes•16m ago
This is the “back to office” of education. It is not a one size fits all solution. There are so many remote and hybrid classes now you guys sound outdated.
analog31•4m ago
That’s fair, but at the same time, expecting any learning to occur in remote classes, when fair evaluation is impossible, may also be outdated.
grahamburger•8m ago
I've heard some teachers are assigning their students to 'grade' a paper written by LLM. The students use an LLM to generate a paper on the topic, print it out, then notate in the margins by hand where it's right and wrong, including checking the sources.
2pEXgD0fZ5cF•7m ago
I don't see the whole AI topic as a large crisis, as others have mentioned: put more emphasis on in-person tests and exams. Make it clear that homework assignments are for practice and learning. If a person thinks that copy/pasting helps them, give them the freedom to so, but if as a result they fail the exams and similar in-person evaluations, then so be it. Let them fail.

I would like to hire students who actually have skills and know their material. Or even better, if AI is actually the amazing learning tool many claim then it should enhance their learning and as a result help them succeed in tests without any AI assistance. If they can't, then clearly AI was a detriment to them and their learning and they lack the ability to think critically about their own abilities.

If everyone is supposed to use AI anyway, why should I ever prefer a candidate who is not able to do anything without AI assistance over someone who can? And if you hold the actual opinion that proper ai-independent knowledge is not required, then why should I hire a student at all instead of buying software solutions from AI companies?

Mobile carriers can get your GPS location

https://an.dywa.ng/carrier-gnss.html
204•cbeuw•2h ago•129 comments

Genode OS is a tool kit for building highly secure special-purpose OS

https://genode.org/about/index
62•doener•2h ago•5 comments

Finland to end "uncontrolled human experiment" with ban on youth social media

https://yle.fi/a/74-20207494
188•Teever•3h ago•129 comments

Demystifying ARM SME to Optimize General Matrix Multiplications

https://arxiv.org/abs/2512.21473
3•matt_d•8m ago•0 comments

Animated AVIF for the Modern Web

https://arthur.pizza/2025/12/animated-avif-for-the-modern-web/
57•sdoering•5d ago•31 comments

CPython Internals Explained

https://github.com/zpoint/CPython-Internals
108•yufiz•4d ago•27 comments

Death Note: L, Anonymity and Eluding Entropy (2011)

https://gwern.net/death-note-anonymity
24•teej•1h ago•4 comments

US has investigated claims WhatsApp chats aren't private

https://www.bloomberg.com/news/articles/2026-01-29/us-has-investigated-claims-that-whatsapp-chats...
11•1vuio0pswjnm7•2h ago•186 comments

Guix System First Impressions as a Nix User

https://nemin.hu/guix.html
100•todsacerdoti•8h ago•39 comments

"Giving up upstream-ing my patches & feel free to pick them up"

https://mail.openjdk.org/pipermail/hotspot-dev/2026-January/118080.html
90•csmantle•9h ago•35 comments

We have ipinfo at home or how to geolocate IPs in your CLI using latency

https://blog.globalping.io/we-have-ipinfo-at-home-or-how-to-geolocate-ips-in-your-cli-using-latency/
167•jimaek•10h ago•44 comments

Nintendo DS code editor and scriptable game engine

https://crl.io/ds-game-engine/
7•Antibabelic•1h ago•1 comments

NASA's WB-57 crash lands at Houston

https://arstechnica.com/space/2026/01/one-of-nasas-three-wb-57-aircraft-just-did-a-belly-landing-...
122•verzali•3d ago•66 comments

My Ridiculously Robust Photo Management System (Immich Edition)

https://jaisenmathai.com/articles/my-ridiculously-robust-photo-management-system-immich-edition/
189•jmathai•3d ago•82 comments

Quaternion Algebras

https://jvoight.github.io/quat.html
82•teleforce•4d ago•34 comments

Berlin: Record harvest sparks mass giveaway of free potatoes

https://www.theguardian.com/world/2026/jan/31/record-harvest-berlin-giveaway-potatoes
20•novaRom•58m ago•12 comments

Show HN: I trained a 9M speech model to fix my Mandarin tones

https://simedw.com/2026/01/31/ear-pronunication-via-ctc/
394•simedw•19h ago•118 comments

Google Cloud suspended my account for 2 years, only automated replies

16•andylizf•1h ago•7 comments

Insane Growth Goldbridge (YC F25) Is Hiring a Forward Deployed Engineer

https://www.ycombinator.com/companies/goldbridge/jobs/78gGEHh-forward-deployed-engineer
1•alvinsalehi•8h ago

What's going on at Heathrow Airport

https://blog.billyedmoore.com/heathrow
16•Billyedmoore•5d ago•7 comments

Antirender: remove the glossy shine on architectural renderings

https://antirender.com/
1756•iambateman•1d ago•420 comments

Show HN: An extensible pub/sub messaging server for edge applications

https://github.com/narwhal-io/narwhal
7•ortuman•3d ago•0 comments

HTTP Cats

https://http.cat/
529•surprisetalk•1d ago•83 comments

Peerweb: Decentralized website hosting via WebTorrent

https://peerweb.lol/
334•dtj1123•23h ago•108 comments

Sumerian Star Map Recorded the Impact of an Asteroid (2024)

https://archaeologyworlds.com/5500-year-old-sumerian-star-map-recorded/
124•griffzhowl•12h ago•43 comments

Students using “humanizer” programs to beat accusations of cheating with AI

https://www.nbcnews.com/tech/internet/college-students-ai-cheating-detectors-humanizers-rcna253878
45•unpredict•3d ago•52 comments

Film students who can no longer sit through films

https://www.theatlantic.com/ideas/2026/01/college-students-movies-attention-span/685812/
61•haunter•3h ago•99 comments

Show HN: Moltbook – A social network for moltbots (clawdbots) to hang out

https://www.moltbook.com/
74•schlichtm•2d ago•783 comments

Apple Platform Security (Jan 2026) [pdf]

https://help.apple.com/pdf/security/en_US/apple-platform-security-guide.pdf
110•pieterr•4h ago•79 comments

Euro firms must ditch Uncle Sam's clouds and go EU-native

https://www.theregister.com/2026/01/30/euro_firms_must_ditch_us/
646•jamesblonde•9h ago•577 comments