frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Ask HN: Scheduling stateful nodes when MMAP makes memory accounting a lie

6•leo_e•2h ago•3 comments

Ask HN: Hearing aid wearers, what's hot?

320•pugworthy•17h ago•187 comments

GhostBin A lightweight pastebin, built with Go and Redis

2•sanaf•2h ago•0 comments

A logging loop in GKE cost me $1,300 in 3 days – 9.2x my actual infrastructure

5•nthypes•4h ago•2 comments

Ask HN: Good resources to learn financial systems engineering?

127•_1tan•1d ago•24 comments

Tell HN: Cursor charged 19 subscriptions, won't refund

3•devtailz•4h ago•2 comments

Tell HN: Declaration of Independence is 100% AI, according to AI Checker

7•whatamidoingyo•3h ago•2 comments

Ask HN: Photos corrupted on Google Pixel phones over time?

5•poolnoodle•7h ago•1 comments

Don't obsess with security and privacy unless they are your core business

6•amano-kenji•13h ago•10 comments

Malicious Bun Script Found in NPM Package Bumps

3•kothariji•10h ago•1 comments

Tell HN: Wanted to Give Dang Appreciation

37•razodactyl•20h ago•3 comments

Ask HN: What tools do you pay for today that feel overpriced or frustrating?

5•psicombinator•16h ago•6 comments

Why isn't There a open-source (project) game?

3•triilman•19h ago•6 comments

ZetaShare Building private file transfer with WebRTC

3•masterdegrees•22h ago•0 comments

Malware in PostHog NPM packages

9•roskoalexey•10h ago•7 comments

Ask HN: Is America in Recession?

20•register•1d ago•27 comments

Ask HN: Advice for feeling like a failure in PhD?

8•phdthrowaway1•1d ago•2 comments

Ask HN: How do you balance creativity, love for the craft, and money?

14•introvertmac•2d ago•9 comments

Boring Laser Eyes Simulator: Add laser beams to your eyes with your webcam

3•frankhsu•2d ago•0 comments

Ask HN: Working in a language that isn't your native one. How hard was it?

8•william-cooke•2d ago•16 comments

Ask HN: Where can you find old NetBSD packages?

12•GaryBluto•2d ago•5 comments

Facebook has made it impossible to delete Pages – dark patterns everywhere

47•ramharts•5d ago•16 comments

Ask HN: First Steps with a Patent Troll?

10•throwawaynvmbr•2d ago•11 comments

Ask HN: Is it time to measure Inflation and CPI without the government?

16•cyrusradfar•2d ago•6 comments

Ask HN: What is the current state of the art in BIG (>5TB) cloud backups?

22•jacobwilliamroy•5d ago•20 comments

Ask HN: Current state of Android USB tethering?

8•namesarehard•2d ago•0 comments

Ask HN: Where could I find early adopters?

4•nonmaskable•23h ago•1 comments

Ask HN: How would you architect a RAG system for 10M+ documents today?

21•Ftrea•3d ago•8 comments

Warp Terminal Doing a Pricing Shock for Canadian Users – 10× Increase on Credits

4•osmosismartini•2d ago•3 comments

Fun weekend task – Calculate your crypto relief or regret

6•shouldabought•3d ago•1 comments
Open in hackernews

Tell HN: Declaration of Independence is 100% AI, according to AI Checker

7•whatamidoingyo•3h ago
My wife is a freelance writer/translator, and so many of her clients check her work with so-called AI Checkers, often making her rewrite her work until it scores below 30% (even if it was 100% her work in the first place).

If you're relying on these tools, please know that they're not accurate. At all. In fact, they're a waste of time. As an example, I pasted the first three paragraphs of the Declaration of Independence into one of these checkers, and the result: 100% AI generated via ChatGPT.

Edit: removed link to AI Checker site, as I don't want this to look like an ad. If you want to check, google "AI Checker", and test yourself.

Comments

WheelsAtLarge•2h ago
I would think so. I'm sure it's one of the documents used to train the LLMs. It does not surprise me that a checker would think it was AI generated. I bet if you put any popular document, the same will happen.

True, people are putting too much trust into the checkers.

dbaman•2h ago
But is that first point really true?

If a document is used to train an LLM, that means that an AI checker would be more likely to give a false positive that the document is written by AI?

I guess it depends on the AI checker.

I would think that that LLM, after generating the fragment "When in the course of.." would just be a bit more likely to generate "human" as the next word, yes?