Eftersom språkmodeller inte är äkta ai utan bara stokastiska ordpapegojor så utvecklar dom samma modulära kognitiva arkitektur som våra hjärnor helt spontant. Här skulle vårt eget AI forums experter kunna göra en stor välgärning om dom kontaktade dom här stackars vilseledda MIT forskarna som är specialiserade på neurokognition och förklara vänligt för dom att dom måste ha fått allt om bakfötterna
Citat:
Modular Cognitive Architecture Emerges
in Large Language Models
Pengrui Han·
Jacob Andreas·
Evelina Fedorenko†·
Andrea Gregor de Varda†
Abstract
The human brain exhibits a striking degree of functional specialization, with distinct networks supporting language, formal reasoning, reasoning about other minds, and reasoning about the physical world. Is this modular organization a fundamental principle of how intelligent systems must be built, or an evolutionary accident specific to biological brains? Here, we test whether a similar organization emerges in Large Language Models, another class of intelligent systems created through a very different optimization process. Using circuit analyses across
N = 46 tasks spanning four cognitive domains (language, formal reasoning, social reasoning, physical reasoning), we find that LLMs develop a modular architecture that mirrors the human brain: tasks drawing on the same network in humans recruit overlapping neurons in LLMs, whereas tasks drawing on different networks recruit distinct neurons. The convergent emergence of modularity in brains and neural networks suggests that it may be a fundamental property of intelligent systems.
https://pengrui-han.github.io/LLM_Modularity_Page/
Kina ska integrera maskin intelligens vertikalt och horisontellt i alla utbildningar så fort som möjligt på order av presidenten. Vad fånigt av dom, här kan forumets experter tjäna mycket pengar om dom bara kontaktar Kina och förklarar missförståndet.
Citat:
Adhering to the principles of putting people first, prioritizing literacy, driving by application, and using AI for good, the plan will unleash AI as an engine for educational transformation. It aims to integrate intelligent technologies across all educational elements, processes, and scenarios. It coordinates AI talent cultivation and application innovation, pushes forward basic infrastructure and innovation ecosystems, and accelerates the creation of a new smart education model that features human-machine collaboration, virtual-real integration, and ubiquitous access.
Citat:
By 2030, a pattern of deep integration between AI and education will basically take shape. An AI education framework spanning all schooling stages alongside a general AI literacy system for the whole society will be built, with both vertical continuity and horizontal connectivity. The scale and quality of AI talent cultivation will significantly improve, and a long-term mechanism for cultivating AI literacy among the entire population will be established.
https://gaj.wuhan.gov.cn/jmzx/gwyxx/202606/t20260630_2814790.html
Det finns böcker som beskriver i utmärkt explicit detalj alla steg som krävs för att bygga en AI. Vad händer om jag tar den här boken som PDF och laddar upp på Mythos och ber den att börja bygga enligt instruktion? Det är vad både Chat och Anthropic planerar att göra internt med sina megakluster. Domhar själva garanterat tusen gånger mer detaljerade loggar kring utvecklingen av sina egna system.
Citat:
Build a Reasoning Model (From Scratch) is a practical guide to understanding how modern reasoning-oriented LLMs work by building their core methods step by step. The book tells a clear engineering story: start with a conventional pre-trained LLM, learn how text generation works, build reliable evaluation tools, improve reasoning through inference-time methods, then move into training-based approaches such as reinforcement learning and distillation.
The progression is deliberate. Early chapters establish the baseline model and explain text generation, KV caching, and evaluation with math verifiers. The middle chapters show how reasoning can be improved without changing model weights, using chain-of-thought prompting, sampling, self-consistency, response scoring, and self-refinement. Later chapters move to changing the model itself through reinforcement learning with verifiable rewards, GRPO improvements, format rewards, and finally distillation from stronger reasoning models into smaller ones.
The book is especially useful because it implements the core methods from scratch rather than treating them as black-box library calls. Readers see how self-consistency, self-refinement, Best-of-N, and training-based methods actually work, including their cost and latency trade-offs. It also discusses common failure modes, including cases where refinement can make answers worse. Difficult concepts such as softmax, temperature, and top-p sampling are clarified with code-linked explanations and diagrams, and visual workflows make pipelines and scoring methods easier to follow.
https://www.manning.com/books/build-a-reasoning-model-from-scratch
Automatiskt arbetsflöde med flera olika frontier modeller som får bolla matematiska problem fram och tillbaka och automatiskt kommer fram till nya insikter inom komplicerade matematiska domäner, formaliserar, summerar, kontrollerar och till slut skriver ett färdigt matematisk uppsats om varje fynd. Just nu måste dom göra så att flera modeller tar det fram och tillbaka, men hur lång tid tar det innan du bara kan prompta en enda modell så gör den allt detta självständigt?
Citat:
Each problem links to its current write-up and, where available, a machine-checked Lean 4 formalization. Write-ups are uploaded gradually and may be merged into a single paper over time.
COLT open-problem track- Shuffled SGD — the SS–RS–GD inequalities (Yun, Sra, Jadbabaie, COLT 2021) — paper
- Learning measured-output quantum circuits (Kun and Reyzin, COLT 2015) - paper (Partial solution).
- Unweighted data selection for linear regression (Hanneke, Moran, Shlimovich, Yehudayoff, COLT 2025) — paper (Problem 3)
- Robust conditional probability estimation (Langford, COLT 2010) — paper
FOCS 2023 (authors' own open question)- Adversarial robustness of online leverage-score sampling (Jiang, Peng, Weinstein, FOCS 2023) — paper
Commutative ring theory — Glaz et al., Open Problems in Commutative Ring Theory- Problem 4: finite-conductor vs. quasi-coherent rings — paper, Lean
- Problem 20 (Cahen–Fontana–Frisch–Glaz) — paper, Lean
- Problem 27: integer-valued polynomials over algebras (Werner) — paper, Lean (27b)
- Problem 30(c) — paper, Lean
Erdős problems- Problem 477: tiling complement — paper
https://github.com/Pengbinghui/pipeline-math