Hacker News: dopamine_daddy

New comment by dopamine_daddy in "Over 30% of new ArXiv submissions now read as AI-written"

dopamine_daddy — Mon, 20 Jul 2026 18:29:48 +0000

You raise a valid point. Just by intuition I'd say if this were true, it would probably just be a small fraction of the actual flagged articles. I will still look into how I can mitigate this when I update the detector.

The difficulty with this is then: How do you get a clean post 2023 dataset? I have no straightforward idea for this. You can't use other AI detectors to build it because then you'd never outperform them.

New comment by dopamine_daddy in "How we measured AI writing across arXiv, and where the measurement breaks"

dopamine_daddy — Mon, 20 Jul 2026 17:50:56 +0000

Yes that is exactly what I suspected. And I see absolutely nothing wrong with this.

New comment by dopamine_daddy in "Over 30% of new ArXiv submissions now read as AI-written"

dopamine_daddy — Mon, 20 Jul 2026 17:48:43 +0000

Thank you, I plan to release the arxiv preprint codes. Also don't worry about killing my server, let me know if you succeed :D

New comment by dopamine_daddy in "Over 30% of new ArXiv submissions now read as AI-written"

dopamine_daddy — Mon, 20 Jul 2026 17:39:44 +0000

Yes, I’ve thought about this too. The strength of these models is that there is a lot more knowledge encoded in them than the average scientist has in mind at any given time. That means they can explore many more possible combinations of concepts.

If we imagine a set of all human ideas that these models have access to, then the set of possible discoveries would be something like the superset of all possible combinations of those ideas. I think all LLM discoveries are bounded by that space.

Looking at the recent OpenAI math discoveries, that seems to be pretty much what happened. Existing ideas were used as building blocks, the model found a valuable combination, and the result was something new that had real value.

New comment by dopamine_daddy in "Over 30% of new ArXiv submissions now read as AI-written"

dopamine_daddy — Mon, 20 Jul 2026 17:26:00 +0000

That's honestly so good to hear, thank you.

New comment by dopamine_daddy in "How we measured AI writing across arXiv, and where the measurement breaks"

dopamine_daddy — Mon, 20 Jul 2026 17:16:00 +0000

We can't know if real science is happening in the background but I'd wager that the majority of these papers is not complete slop but real findings with AI generated text used to communicate it. If it was just straight slop I would be really worried.

New comment by dopamine_daddy in "Over 30% of new ArXiv submissions now read as AI-written"

dopamine_daddy — Mon, 20 Jul 2026 17:14:25 +0000

Yeah I was careful on purpose with my statement. :D

New comment by dopamine_daddy in "How we measured AI writing across arXiv, and where the measurement breaks"

dopamine_daddy — Mon, 20 Jul 2026 17:11:34 +0000

I tried my best to avoid leakage. If you're curious about how I trained the detector I have a writeup on it: https://unslop.run/blog/how-our-ai-text-detector-works

FYI this is all relatively new so there might be lots of issues and iterations coming.

New comment by dopamine_daddy in "How we measured AI writing across arXiv, and where the measurement breaks"

dopamine_daddy — Mon, 20 Jul 2026 17:08:31 +0000

Surprisingly I agree with you. My opinion is: if it makes communicating research more effective, while not reducing the quality of the output substantially, I see no issue.

A possible conclusion for this could be: If the majority of CS papers is AI written, let's just accept this reality universally and stop worrying about it altogether.

New comment by dopamine_daddy in "How we measured AI writing across arXiv, and where the measurement breaks"

dopamine_daddy — Mon, 20 Jul 2026 16:36:36 +0000

I scored the full text of 12,750 arXiv papers from 2021 through 2026 to find out how many of these get flagged as machine written and how much it increased since the release of chatGPT. I purposely tuned the detector to avoid false positives. My detection rate pre chatGPT is around .4% for that reason.

The biggest results: in Jan of 2026 about 39% of papers got flagged as AI written. In computer science speicifcally the peak was at 65%. Mathematics barely moved away from 0.7%, though the proof heavy math texts might just not get picked up by the detector properly.

All this is a detector estimate of a statistical signal and not a proof any given author used AI. Machine written can also mean heavy AI-assisted editing.

How we measured AI writing across arXiv, and where the measurement breaks

dopamine_daddy — Mon, 20 Jul 2026 16:36:36 +0000

Article URL: https://unslop.run/blog/measuring-ai-writing-on-arxiv

Comments URL: https://news.ycombinator.com/item?id=48981206

Points: 242

# Comments: 168

New comment by dopamine_daddy in "Slople – can you pass the reverse Turing test?"

dopamine_daddy — Sat, 18 Jul 2026 11:13:58 +0000

I made a game where you rewrite a sentence to try and make it sound like it was written by an LLM. It scores you with my own mixture-of-experts MoE AI-text detector. I honestly built it to see what creative ways people come up with to fool it. It's probably not that good right now, except on academic writing, which is mostly what I trained it on.

Slople – can you pass the reverse Turing test?

dopamine_daddy — Sat, 18 Jul 2026 11:13:58 +0000

Article URL: https://unslop.run/slople

Comments URL: https://news.ycombinator.com/item?id=48956958

Points: 1

# Comments: 1