Daniel Nicorici retweeted
Gemini Argon sounds so much more natural than Claude or GPT.
Pangram agrees: On a 1,680 words Argon-generated document in which I changed two words only, Pangram gives a 48% human rating.
Daniel Nicorici retweeted
There is an unusual anomaly with narcissists/psychopaths and sex.
First, the data shows that predators/dark personalities are completely boundaryless regards sex and relationships. There is no place they will not go - everything is on the table. They may engage coercion, dishonesty, violence. They will break laws, codes, agreements and moral boundaries. They may use sex to control, to leverage, to manipulate and to win.
The anomaly is that because they don’t experience emotions due to brain differences, some predators appear as asexual. Predators do not experience love so for them sex is an exploration, a sensation, a tool, a means to control. Love and intimacy is not a part of it although they can act loving.
Some choose not to act and are completely cold around sexuality and intimacy.
Daniel Nicorici retweeted
Not to mention a lot of the publications produced provide negative information, such that better decisions would be made if the work were not done at all
“We need less research, better research, and research done for the right reasons” - Altman
Daniel Nicorici retweeted
This is a slide from Terence Tao's 'Math 2.0' public lecture yesterday evening at Caltech. I am highly confident that, if the cure actually worked, almost any Stage 4 cancer patient would tell you that human comprehension of its mechanism is completely irrelevant to them.
Daniel Nicorici retweeted
1913: Ramanujan sends Hardy theorems with no proofs, no chain of thought. Hardy: “Bring this genius to Cambridge!”
2026: AI produces mathematical proofs. Mathematicians: “This threatens the integrity of our field!”
Daniel Nicorici retweeted
"Mathematicians did not ask for maths to be solved."
In Germany we call these people "Fachidioten".
It means they put all their skill points into one subject and are fucking retarded everywhere else.
What do we get next? Cancer researchers complaining AI solved cancer?
This quoted post is unavailable.
Daniel Nicorici retweeted
Scientific discoveries do not require prior permission
Daniel Nicorici retweeted
My partner is a theoretical computer scientist (essentially a mathematician). She has mixed feelings about AI's impact on math research overall, but her feeling about today's release is one of overwhelming awe and excitement. Literal tears of joy. Quote: "so lucky to be alive rn"
OpenAI is downplaying this for PR reasons. If you’re a mathematician, you must feel like a nuclear bomb hit, and you’re at ground zero.
Reasoning models are two years-old. In that time they went from incapable of basic arithmetic to solving problems humans couldn’t solve for decades.
Math is only the beginning. AI will revolutionize the entirety of the human scientific endeavor. Most of us don’t appreciate what that means.
What a time to be alive!
Daniel Nicorici retweeted
I find it outrageous that frontier closed-weight LLMs inject statistics-based watermarks into the texts they generate for me.
I paid for this text; it belongs to me, and I don't want it to be marked in any way.
Furthermore, I'm 100% sure that such texts, when detected on a webpage, will decrease their relevance in web search results or result in an entire website shadowban by search engines.
So, to get rid of the watermarks on the AI-generated texts used on ChapterPal, I developed a watermark removal tool I call Dehydrate.
Watermark technologies such as SynthID used by Google and Anthropic and textGrain by OpenAI embed statistical patterns in the words used in these texts.
Dehydrate modifies these patterns by moving or replacing some words to make watermark recognition unreliable while preserving the meaning of the text and its individual sentences and paragraphs.
If you want to try it for free, you need a ChapterPal account: chapterpal.com/dehydrate
It's a quickly-made UI, so if you find any bugs, please let me know.
Daniel Nicorici retweeted
Referring to hundreds of solutions to important and difficult mathematics problems, some of them central to entire fields, and many with lean formalizations as "slop" (low quality, mass produced, digital junk) is unserious.
Daniel Nicorici retweeted
🧵UChicago is leading the way on AI in higher education.
1. All students will take an analog writing class.
2. Core classes in the humanities and social sciences will prohibit technology in class.
3. A new requirement in computational thinking will likely be added to the core.
Daniel Nicorici retweeted
Are you brave enough to ask ChatGPT Astra to look for bugs in all your pre-LLM papers?
Daniel Nicorici retweeted
ok, so I had codex check this out and broadly Emulate checks out.
Codex had emulate generate 100 texts ran each through pangram
result? 85 were labeled Human, 7 were "mixed" and 8 were AI.
one downside seems to be writing quality seems to suffer as measured by adherence, coherence and usefulness.
all data and code here:
github.com/dalek2point3/emul…
Daniel Nicorici retweeted
A bold and I think brilliant essay by philosopher Joe Cruz at Williams College.
We can’t stop students from using AI. As writing can no longer certify intelligence, we'll have to privilege its IRL performance.
Daniel Nicorici retweeted
We've expanded the checks that paperpush performs prior to submission. We're finding it very useful as a final QC prior to submitting work:
paperpush.readthedocs.io/en/…
Daniel Nicorici retweeted
I'd recommend that clinicians begin with principles of study design and measurement, and then spend time designing a hypothetical perfect RCT. Then learn how best to approximate the perfect design/perfect measurements.
Daniel Nicorici retweeted
Check our new @biorxivpreprint preprint on a method for masking low-complexity regions in genome databases, led by Yuchin (Peter) Ge in my lab. Also includes a new release of a very large, pre-masked microbial database: biorxiv.org/content/10.64898…
Daniel Nicorici retweeted
Dickens: 1859.
Pangram: 65% AI.
AI detectors don’t detect authorship. They detect patterns associated with AI-generated text.
That distinction matters—especially when a percentage becomes an accusation.
Daniel Nicorici retweeted
A great example of what you are asking for: onlinelibrary.wiley.com/doi/… (1995!!) . Not only are many AI tools a rebranding of statistics, but AI "experts" often manage to inject statistical methodology errors in the process.
