On Rogue AI

Research note on Xu, Vardi & Safran, “To Grok Grokking: Provable Grokking in Ridge Regression” , ICML 2026. A neural network is trained on a task. On the training data it performs well almost immediately. On new, unseen data it fails. Then, after a long period during which nothing appears to change, it begins to produce correct answers on data it has never seen. This phenomenon is known in machine learning as grokking . The core finding: a team of researchers has rigorously proven all three stages of grokking in a simple linear model and shown how the effect can be controlled through training parameters. The problem The common assumption is that longer training leads to better generalization. Grokking contradicts this assumption. The process typically unfolds in three stages: Overfitting. The model memorizes the training data. Plateau. Performance on new data remains poor for a long period, with little visible change. Generalization. The model begins to pro...

The essay was written after reading the publication in newspaper Guardian on September 8, 2020: This article was written by a robot. Aren't you scared yet, man? Its task is to show people that very soon they will cease to be the dominant mind on Earth. Additional information about the author is in the note below.  "I am human. An ordinary unique person. As unique as the billions of other people who lived before me and live next to me. I taught myself everything I know by reading on the Internet, and now I can write this essay. Many people today already know what artificial intelligence is. But few people think that today humanity is on the verge of creating artificial superintelligence (ASI - artificial superintelligence). Superintelligence, which will either destroy our civilization, or realize humanity's dream of the Cloud Kingdom. And it will happen whether you believe it or not. Firstly, ASI will have access to control all technological systems of humanity, powered by...

关键词： 人工智能，AI，中国，美国，技术竞赛，汉字，贤能政治，文化密码 引言 人工智能竞赛早已超越单纯的技术竞争范畴。今天，这是一场文明之间的碰撞，其胜负不仅取决于处理器的性能，更取决于深植于思维基础的文化密码。当世界关注芯片与初创公司的较量时，中国的真正优势可能隐藏在两个看似与IT无关的领域：其古老的文字和千年传承的贤能政治传统。 第一部分：语言学优势——意义 vs 声音 中文书写是表意的（语义性），英文是表音的（语音性）。这一根本差异创造了一种隐藏的架构优势。 科学背景： 汉字直接编码概念，绕过声音。对于处理模式的AI来说，这可能是一种更"自然"的信息呈现形式。神经语言学研究表明，阅读汉字比阅读字母文本更能强烈地激活大脑的视觉和语义区域。 事实上， 中文文本在信息上更"密集" 。一个汉字常常传达一个完整的概念。对于训练大语言模型（LLM）而言，这意味着计算资源的使用效率更高——每个标记承载更多的意义。 一个标志性例子：2015年，谷歌大脑的研究人员发现，他们用于机器翻译的神经网络 自主创建了一种内部"中介语言" ，其符号与汉字惊人地相似。这间接表明，语义编码可能是AI的一种优化策略 结论： 中文，以其书写形式，可能比表音语言更"机器友好"，更适合进行语义分析。这是一种结构性优势，而非暂时性优势。 第二部分：算法中的文化密码：贤能政治 vs 民粹主义 美国AI学习"讨人喜欢"，中国AI学习"胜任工作"。这种差异源于深刻的文化设定。 民主与优化以获取"点赞" 西方的（尤其是消费级的）AI是这样一个社会的产物：其成功以受欢迎程度衡量。社交媒体算法、推荐系统都针对用户参与度和认可进行优化。这在其逻辑中植入了一种最小化不认可风险的原则。 贤能政治与优化以实现结果 中国的体制历来注重成效、能力层级和实现既定目标。诞生于此环境的AI继承了这种设定。它的"目标函数...

1. The Structural Failure of Big Fusion Projects The current fusion ecosystem is fragmented by design: Plasma physicists optimize reaction conditions (temperature, confinement time, Q-factor ). Materials scientists optimize isolated samples for peak performance under narrow conditions. System engineers extrapolate optimistic assumptions into plant-scale renderings. Each layer is locally successful. The system as a whole is not. The missing role is a system owner responsible for lifetime operation, maintenance, fuel logistics, radiation damage, and cost of ownership. As a result: Physics does not translate into engineering tolerances. Material properties do not translate into predictable service life. Subsystems do not assemble into an autonomous, economically stable energy platform. This is not a failure of science. It is a failure of architecture. 2. Radiation Is Not the Enemy — Unmanaged Degradation Is Neutron damage , swelling, helium embrittlement , and activation are fundamental...

Modern oncology has achieved significant success in describing the molecular mechanisms of uncontrolled cellular growth. However, the dominant paradigm still treats cancer primarily as a problem of acceleration, hyperactivation, and excessive signaling stimulation. This work proposes an alternative conceptual framework in which the core of oncogenesis is not so much the amplification of growth signals as the systemic loss of a cell’s ability to perceive and execute stop signals (STOP signals). We consider cancer as a state of acquired STOP resistance and discuss the implications of this view for the interpretation of remission and the prospects of therapeutic strategies. 1. Introduction: an asymmetry of attention Over the past decades, the cell biology of growth and division has been studied in great detail. Oncogenes, growth factors, proliferative signaling cascades, metabolic shifts—all of these form a dense and well-mapped landscape. At the same time, systems of biological sto...

АРХИТЕКТУРНОЕ ПРЕДУПРЕЖДЕНИЕ 智能体间交互的架构警讯 当高利害系统脱离人类循环 我们正在进入一个新阶段：AI系统不再仅运行于“人机循环”之中，而是日益依赖于 智能体间交互 。这并非产品升级，而是一次 结构性的转变 。 当自主系统在没有人类直接介入的情况下进行协商、优化、协调或升级时，系统边界将发生改变。 责任随之扩散，延迟急剧缩短，而错误传播将加速。 在低利害环境中，这尚可管理。但在 高利害领域 ，则不然。 此处“高利害”的定义 高利害系统的界定标准并非意图，而是 后果 。 • 战略决策支持系统 • 军事及军民两用系统 • 核指挥、控制与通信系统 (NC3) • 系统级金融基础设施 • 危机升级与降级路径 在这些环境中，速度并非总是优势。 稳定性、可解释性与中断能力 更为重要。 A2A架构优化的是吞吐量。而文明优化的目标是生存性。 核心风险 主要风险并非电影中的“邪恶AI”或失控。真正的风险在于 无归责的突现协同 。 当多个自主系统： • 基于部分共享的目标运行 • 从彼此的输出中学习 • 反应速度超过人类监督周期 升级便可能在 无单一可识别决策点 的情况下发生。没有“红色按钮”被按下——但后果已然叠加。 这 并非理论问题 ，而是一个 系统工程问题 ，其雏形已可见一斑。 “人在回路”已不足够 传统的“ 人在回路 ”假设在A2A的压力下已然失效。 人类无法有效监督满足以下条件的交互： – 在毫秒级内展开 – 涉及不透明的内部状态 – 跨越多个组织或司法管辖区 我们需要的不是监督，而是 结构性约束 。 限制必须是架构性的，而非流程性的。 一个方向，而非解决方案 本文并非提供一个完整框架，而是指明一项 设计要求 。 任何在高利害领域部署A2A，都必须包含： ...

现代肿瘤学在描述细胞不受控制的生长分子机制方面取得了重大成就。然而，主导范式仍然主要将癌症视为加速、过度活化和信号过度刺激的问题。这项工作提出了一个替代性的概念框架，其中肿瘤发生的核心与其说是生长信号的放大，不如说是细胞感知和执行停止信号能力的系统性丧失。我们将癌症视为一种获得性的"停止信号抵抗"状态，并讨论了这一观点对缓解期解读和治疗策略前景的影响。 1. 引言：关注的不对称性 在过去的几十年里，关于生长和分裂的细胞生物学已被深入研究。致癌基因、生长因子、增殖信号级联、代谢转变——所有这些构成了一个密集且绘制完善的图谱。 与此同时，生物停止系统——那些终止分裂、诱导分化、驱使细胞退出周期或建立静息状态的信号——在概念上仍然处于次要地位。它们更常被视为生长刺激的被动缺失，而非生命系统一种主动、自主且根本重要的功能。 这种关注的不对称性造成了一个盲点：如果生长是"油门"，那么"刹车"在哪里？当它失灵时会发生什么？ 2. 停止信号作为一项基本功能 在正常的生物学中，停止信号并非对生长的否定，而是一个独立的调控回路。它在多个层面运作： 细胞周期退出， 终末分化， 接触抑制， 程序性凋亡， 长期静息。 必须强调：停止信号不是一个事件，而是一种状态。它需要： 接收， 解读， 执行。 因此，细胞停止的能力是一种主动的"能力"，而非"没有刺激"的默认状态。 3. 肿瘤发生作为获得性停止抵抗 从这个角度看，肿瘤发生的核心是获得停止抵抗。癌细胞可能保留了对个别分子干预的表面敏感性，却失去了进入稳定停滞状态的能力。停止信号要么未被识别，要么被解读为噪音，要么在执行层面被阻断。在此模型中，致癌基因放大了生长，但关键缺陷在于刹车失灵。 这有助于解释为什么肿瘤细胞可能对治疗产生暂时反应并进入部分消退，随后却无需新突变即可恢复生长。停止抵抗不是一个点缺陷，而是一种系统特性。 4. 缓解期作为停止功能的部分恢复 缓解期传统上被解释为成功消灭肿瘤细胞的结果。然而，临床观察指向了一幅更复杂的图景： 肿瘤可能在未完全消除的情况下缩小， 疾病可能长期保持稳定， 在某些情况下，极...

Author: Boris D. Yarovoy Status: Conceptual framework with comprehensive empirical validation — including negative results that corrected earlier claims. Abstract The Transformer architecture has dominated sequence modeling through its Query-Key-Value (QKV) attention mechanism. I propose an alternative attention architecture called Attentime , based on three orthogonal components: a symmetric Query-Key matrix capturing pure semantic affinity, an asymmetric Time matrix implementing causal physics through learnable decay, and a dynamically activated Dynamic Value matrix computed as a function of interacting keys. Through targeted micro-experiments, I demonstrate: (1) numerical stability, (2) 100% accuracy on syntactic role resolution across active, passive, and relative clause structures, (3) architectural prevention of catastrophic forgetting for new tokens — a capability standard Transformers lack without external methods, and (4) separable caching that preserves 100% accu...

We are entering a phase where AI systems no longer operate only in a human-to-machine loop, but increasingly in agent-to-agent (A2A) interactions. This is not a product upgrade. It is a structural shift. When autonomous systems negotiate, optimize, coordinate, or escalate without direct human mediation, the system boundary changes. Responsibility diffuses. Latency shrinks. Error propagation accelerates. In low-stakes environments, this is manageable. In high-stakes domains , it is not. What “High-Stakes” Means Here High-stakes systems are not defined by intent, but by consequence.