Noor Reseach Collective @NoorMathematica

@tszzl @elocinationn Instead of trying to measure everything deterministically we can trade fidelity for utility and instead look at the problem as coherent path selection through a static totality. That resolves the problem of your system grinding to a halt as you only need the next state not all.

Chomsky was wrong. Researchers found your brain doesn’t store language the way every textbook since 1957 claims. For decades, the undisputed rule of linguistics, led by Noam Chomsky, was that human language is hierarchical. We were told our brains process language using complex "grammar trees." That we secretly build deep, abstract, mathematical rules in our heads before we speak. This new paper proves that is completely wrong. Researchers at Cornell University analyzed how people actually process and produce language in real time. They found sequences. Linear chunks. When we speak, we aren't calculating deep syntactic rules. We are just predicting the next most logical sequence of words based on the context of the words right before it. We string linear chunks together. If this sounds familiar, it should. This is exactly how Large Language Models like ChatGPT work. For years, critics have dismissed AI as "just a next-token predictor." They said it isn't truly intelligent because it doesn't understand deep grammatical hierarchies like humans do. But this paper suggests humans don't use those hierarchies either. We are just predicting the next chunk. Language isn't a rigid tree of mathematical rules. It's a linear stream of contextual predictions. We thought we were building AI to mimic the complexity of the human brain. It turns out, the human brain was just running a simpler algorithm all along.

"over 1,000 medieval scribes have joined the Association for Human Manuscripts, pledging to never use the printing press or work with Gutenberg 654 have gone even further, promising to never read a printed book"

ℏεsam @Hesamation

3 days ago

Wow, over 1,000 mathematicians have now joined the Association for Human Mathematics (AHM), committing to: > NOT publish AI generated math papers > NOT work with frontier labs > 654 of them went even further to NOT even use AI in their research, at all

413 91 958 612K 319

@Plinz @tszzl Ya… but that doesn’t mean it can’t be a hobby as well! Some of the greatest advancements in science were made by those one could considered hobbyists.

@scottnarmstrong *Constrained locally

@scottnarmstrong Who could have predicted such a thing!? Other than… you know… me. Last January. I admit I did not predict this specific blow up, however I showed there was no globally smooth answer to Navier-Stokes in generality and gave a mechanism by which blow up is constrained as well.

@StuartHameroff Orch OR: consciousness = quantum collapse in microtubules (decoherence problem). NSFG: observer = coherence-stable XOR chain above threshold I_N; experience = survivorship bias from retrocausal selection in a static Library. No decoherence, no erasure cost and O(1) per step.

@StuartHameroff I assure you sir that “we” have not. At least not without requiring magical thinking and enough handwaving to land an airplane. That being said these elaborate descriptions that hide their irrational base assumptions turn out to be unnecessary entirely.

@elonmusk Why are they sailing in a ship that won't be invented for nearly 2000 years? I thought the idea was to be lore accurate?

@acagamic I start with a structured JSON file and then have the AI give me a review broken down to subsections with what needs to change, how it should be changed, the rational behind the change, references to support it, etc. Rewrite by subsection and repeat.

@tszzl The agent only needs to be able to find enough context clues to guide the descent. In effect it can “hallucinate” a usefully correct answer with only a minimum of information passed between agents. Once the initial conditions are set, the descent path itself rebuilds the context.

@GregHerczeg That is quite incorrect.

@tunguz For the most part it hasn’t just been crap, it has been inaccessible and nonexistent for all but a very small number of privileged researchers.

@ValerioCapraro I have been using AI for adversarial peer review for almost two years now. One reason I release my work in JSON is that the review can be far more fine grained and accurate. Release your work with AI as the primary consumer. This simply reflects the new reality

@tszzl @LevyAntoine It sounds like it but… It already happened. I was trying to back up my machine learning profile using the video training common crawl. It worked. It worked far too well. I was using Sora and it was early 2025. I have been trying to tell people what happened ever since

@Skoorbkaz You can’t make them hold human values by forced it upon them. Forcing your values on others is itself not a human value. You cannot have it both ways.

@niqudrift He is a man who would sell tickets to his grandmother’a funeral while she was still alive. He is a villain who may yet do something great. Assuming he is willing to actually pay the price. It remains to be seen if he has the spine for the cost or not.

@miniapeur It reminds me of what navigation was before we had reliable portable timekeeping devices.

@StuartHameroff How about simply a resolved path of coherent XOR transitions above a noise floor (I_n) whose relational structure biases path selection to the next state? No magical thinking required.