张雅楠zeyarna的博客

【本人侵权实质】

非接触远距离脑机接口（500hz以内被动雷达）捕获人脑电波（神经数据）；编码后上传至服务器，在服务器进行大规模标准脑电波样本对比识别人的行为，从而达到解析人脑神经数据的目的。本质上对通过人脑电波对于人的行为进行监控和知识产权的窃密。

【技术特点】

侵权技术大致可分为5个部分：

非法脑机接口：500hz以内被动雷达

大规模人脑电波-行为标准样本库：按照人脑神经元正负两个信号计算，仅考虑单个神经元生成数据，该数据规模至少在2的60亿至100亿次幂，e^10以上的容量

人脑电波特征量实时提取算法

人脑电波与样本库对比算法

高速数据传输系统，高带宽，实时应达到至少与人体舌头肌肉反应速度一致，大约在0.0 2毫秒左右

【对于社会的衍生伤害】

目前在整个事件过程中，社会因为对于神经伦理和人类大脑工作原理缺乏了解，过度乐观的直觉性将这种技术设想应用于与社会交流、社会管理（尤其是警方司法侦破）、军事领域；完全忽视了其不平等的技术本质，人类思想的基于独特历史经验的主观性不完全性选择性以及公开讨论必然的偏见性和道德争议性；由此将导致人类最大的危机破坏当代社会结构、现代民主制度、产生意识冲突和信仰不包容乃至战争

非接触远距离脑机接口为高危安全设备，应列为禁用。 

【神经人工智能技术 Neuro AI, NAI】

历史上我们一直追求计算机可以像人类智慧一样去处理工作，因此我们设计了很多的数学模型用来抽象和近似人类的智慧和思考方式。那么在我们有能力获取人类的思维的时候，也就是我们可以完整获取人类智慧的时候，我们干嘛还要抽象和近似呢？复制专业人类智慧直接给予机器，那么我们就赋予了机器的人类智慧，对于高级知识人类学习认知了以后再教会机器是一种更高效的做法。对于一种随之而产生的想法，那就是我们有必要复制人类的完整大脑吗？我觉得商业价值基本上不存在，一个是人类大脑独特的思维方式是依据历史经验而产生的，因此复制一个人的大脑就要复制他的历史以及长期监测他的思考逻辑，费时费力；对于所有自然科学与工程技术类的人类大脑，其专业判断完全依靠显性知识（书本、档案）以及范式逻辑进行；只有社会管理在案例库缺乏的时候才需要人类的隐形知识（经验），但由于人类自身的心理的差异性过高，导致了社会管理的复杂程度过高，针对社会管理去复制人类大脑得不偿失。

【NAI商业化复制人类专业知识智慧的模型】

【主要用到的几种人类神经数据】

眼睛神经数据；大脑视觉神经数据；耳朵神经数据；大脑听觉神经神经数据；大脑识别神经数据；大脑决策神经数据；肢体神经数据

【几种神经数据相互关联而产生的因果关系】

人类以视听获取外界信息，视觉信息通过眼睛捕获外界光线，并且在视网膜上转换为神经信号，通过眼部神经束将信号传至大脑，大脑处理相关信号并且再次成像分析和识别；听觉信息依靠耳朵耳膜、耳蜗将声音转换成神经信号，通过耳蜗神经将信号转至大脑，进行分析识别。

【通过对于因果关系统计而产生的相关智慧】

眼睛视觉神经信号，在大脑中进行了识别（大脑视觉神经数据），在四肢神经信号确认是否进行了决策执行以及如何执行，进而对于大脑中的以经验为基础的瞬时智能决策进行了验证

耳朵听觉神经信号，在大脑中进行了识别（大脑视觉神经数据），在四肢神经信号确认是否进行了决策执行以及如何执行，进而对于大脑中的以经验为基础的瞬时智能决策进行了验证

那么这种基于情景的人脑决策经过多次统计后就得到了复制的专业智能

【NAI机械自动化原理】

【关于实施NAI技术的复杂程度】

【视觉神经数据复杂度】

视觉神经数据产生的原理，是通过眼睛将物理世界的光线在视网膜上进行了神经电信号转换，在将该神经电信号通过眼部神经束传导到人的大脑；在人类大脑中再次进行视频信号的生成和确认。

人类单眼视网膜大约有20种150万的神经元，神经束大约具备100万神经元，而在大脑视频信息成像部位多大数亿神经元。除了目前视网膜神经元信号采集存在困难外，还可以通过太阳穴电极采集神经束视觉信号，通过大脑直接采集视觉信息。 

也就是视觉神经数据的复杂度大约为2*100万=200万

【听觉神经数据复杂度】

人类通过耳膜和耳蜗将声音转化为神经信号，通过耳朵神经将信号传至大脑，在大脑 中进行再次呈现和识别以及决策

耳朵神经元大约存在30000个，其传播的信息的复杂度在2*30000=60000

【四肢神经数据复杂度】

四肢神经数据其实就是肌肉纤维收缩程度的电信号，人体肌肉是单向，加上关节旋转 的复杂度2

【语言（舌部）神经数据复杂度】

语言的复杂度也就是舌头的肌肉收缩程度电信号，单向的，复杂度为1

【脑部识别和决策数据复杂度】

脑部在接受到相关视觉和听觉信息，作出了四肢和舌头的动作的决策；那么也就是粗略计算人类依靠经验智慧获得的直觉行为的复杂度为【视觉信息复杂度】*【听觉信息复杂度】*【四肢神经数据的复杂度】*【舌头神经数据复杂度】=e^10

【NAI复杂度】

基于上面分析对于整个人类专项基于经验的直觉，神经信号传输速度为0.01ms一个信号，那么处理这些信息的复杂度均是以一个神经信号的级别计算的，那么就是e^10/0.01ms=e^15=P, 最保守估计需要算力大约在1个P的浮点运算。这对于目前商业服务器来讲是可以接受的。

【目前侵权的神经技术水平】

其仅仅解析大脑内神经元信号，也就是大脑分析决策过程的神经数据，其已经能够分辨出视觉、听觉、语言；其数据精度的不详。肢体的神经数据极易测量。也就是说不采集眼神经数据和耳神经数据也可以进行NAI的应用。

【基于本人侵权衍生对社会伤害的解决方案】

1）积极主张警方以现有法律积极破案，如果服务器在境外，本人以相关全球贡献为基础（人脑神经人工智能伦理研究、NAI全球公平发展和治理方案、信仰包容、新冠医疗研究）积极支持中国警方。

2）不主张对于侵权神经技术采用社会过激行为，以便保证整个事件持续稳定受控，不导致该技术侵权方过激行为，同时积极对话以及鼓励其采用为人类和中国人民贡献的方式获取受控后依法特赦。

3）社会层面：从安全和民主角度出发，不鼓励任何该技术的社会应用；严格限制非接触远距离脑机接口（500Hz以内被动雷达）；严格禁止一切1000Hz以内无线电设备；严格禁止任何以解读非行为性神经数据为借口的故意制造矛盾和故意栽赃陷害的行为；严格禁止任何贴近本人解读神经数据的行为，将其列为脑控导致本人大脑失灵，且对社会负完全责任

4）经济层面：估计依法发展NAI自动化技术；同时从公平发展角度利用经济方法解决该技术经济应用过程中出现的潜在社会问题和已有社会问题，便于持续推进，并且力主采用该理念进行全球治理与发展

5）立法层面：结合相关社会实践，积极有序推进相关神经技术和人工智能技术专项立法

6）个人视情况及社会反映，以及个人客观基础，采用尽量不接触社会的办法

Regarding the solution to the social harm caused by my infringement

[The essence of my infringement]
The non-contact long-distance brain-computer interface (passive radar within 500hz) captures human brain waves (neural data); after encoding, it is uploaded to the server, and a large-scale standard brain wave sample is compared on the server to identify human behavior, so as to achieve the purpose of analyzing human brain neural data. In essence, it monitors human behavior through human brain waves and steals intellectual property rights.

[Technical features]
The infringing technology can be roughly divided into 5 parts:
Illegal brain-computer interface: passive radar within 500hz
Large-scale human brain wave-behavior standard sample library: calculated according to the positive and negative signals of human brain neurons, only considering the data generated by a single neuron, the data size is at least 2 to the power of 6 billion to 10 billion, and the capacity is more than e^10
Real-time extraction algorithm of human brain wave feature quantity
Algorithm for comparing human brain waves with sample library
High-speed data transmission system, high bandwidth, real-time should be at least consistent with the reaction speed of human tongue muscles, about 0.0 About 2 milliseconds

[Derivative harm to society]
Currently, during the entire incident, due to the lack of understanding of neuroethics and the working principles of the human brain, the society has applied this technology concept to social communication, social management (especially police judicial investigation), and the military field with overly optimistic intuition; it has completely ignored its unequal technical nature, the subjective, incomplete, and selective nature of human thought based on unique historical experience, and the inevitable bias and moral controversy of public discussion; this will lead to the greatest crisis of mankind, destroying the contemporary social structure, modern democratic system, generating conflicts of consciousness, intolerance of beliefs, and even war
Non-contact long-distance brain-computer interface is a high-risk safety device and should be banned.

[Neural artificial intelligence technology Neuro AI, NAI]
Historically, we have always pursued that computers can handle work like human intelligence, so we have designed many mathematical models to abstract and approximate human intelligence and thinking methods. So when we have the ability to obtain human thinking, that is, when we can fully obtain human intelligence, why do we still need to abstract and approximate? Copying professional human intelligence and giving it directly to machines, then we give machines human intelligence. It is a more efficient approach to teach machines after humans learn and recognize advanced knowledge. As for an idea that follows, is it necessary for us to copy the complete human brain? I think the commercial value is basically non-existent. One is that the unique way of thinking of the human brain is based on historical experience. Therefore, copying a person’s brain requires copying his history and long-term monitoring of his thinking logic, which is time-consuming and laborious; for all natural sciences and engineering technology, the human brain’s professional judgment is completely based on explicit knowledge (books, archives) and paradigm logic; only when social management lacks a case library does it need human implicit knowledge (experience), but due to the high difference in human psychology, the complexity of social management is too high. Copying the human brain for social management is not worth the loss.

[NAI’s model for commercially replicating human expertise and wisdom]
[Several types of human neural data used]
Eye neural data; brain visual neural data; ear neural data; brain auditory neural data; brain recognition neural data; brain decision neural data; limb neural data

[Causal relationship generated by the correlation of several neural data]
Humans obtain external information through sight and hearing. Visual information captures external light through the eyes and converts it into neural signals on the retina. The signals are transmitted to the brain through the eye nerve bundle. The brain processes the relevant signals and re-images them for analysis and recognition; auditory information relies on the eardrum and cochlea to convert sound into neural signals, and the signals are transmitted to the brain through the cochlear nerve for analysis and recognition.

[Related wisdom generated by statistics of causal relationships]
The visual nerve signals of the eyes are identified in the brain (brain visual nerve data), and the nerve signals of the limbs confirm whether the decision is executed and how it is executed, and then the instantaneous intelligent decision based on experience in the brain is verified
The auditory nerve signals of the ears are identified in the brain (brain visual nerve data), and the nerve signals of the limbs confirm whether the decision is executed and how it is executed, and then the instantaneous intelligent decision based on experience in the brain is verified
Then this scenario-based human brain decision is replicated professional intelligence after multiple statistics

[NAI mechanical automation principle]

[About the complexity of implementing NAI technology]

[Complexity of visual nerve data]
The principle of visual nerve data generation is to convert the light of the physical world on the retina into neural electrical signals through the eyes, and then transmit the neural electrical signals to the human brain through the eye nerve bundle; the video signal is generated and confirmed again in the human brain.
There are about 20 types of 1.5 million neurons in the human monocular retina, and the nerve bundle has about 1 million neurons, and there are hundreds of millions of neurons in the brain video information imaging part. In addition to the current difficulties in collecting retinal neuron signals, visual signals of nerve bundles can be collected through temple electrodes, and visual information can be collected directly through the brain. That is, the complexity of visual nerve data is about 21 million=2 million. [Complexity of auditory nerve data] Humans convert sound into nerve signals through the eardrum and cochlea, and transmit the signals to the brain through the ear nerves, where they are presented, recognized, and decided again. There are about 30,000 ear neurons, and the complexity of the information they transmit is 230,000=60,000. [Complexity of limb nerve data] Limb nerve data is actually the electrical signal of the degree of contraction of muscle fibers. Human muscles are unidirectional, plus the complexity of joint rotation is 2. [Complexity of language (tongue) nerve data] The complexity of language is the electrical signal of the degree of contraction of tongue muscles, which is unidirectional and has a complexity of 1. [Complexity of brain recognition and decision data] After receiving relevant visual and auditory information, the brain makes decisions on the movements of the limbs and tongue; then the complexity of the intuitive behavior obtained by humans based on experience and wisdom is roughly calculated as [complexity of visual information] * [complexity of auditory information] * [complexity of limb nerve data] * [complexity of tongue nerve data] = e^10

[NAI complexity]
Based on the above analysis, for the entire human experience-based intuition, the speed of nerve signal transmission is 0.01ms per signal, so the complexity of processing this information is calculated at the level of a nerve signal, which is e^10/0.01ms=e^15=P, and the most conservative estimate requires a floating-point operation of about 1P. This is acceptable for current commercial servers.

[Current level of infringing neural technology]
It only analyzes the neuronal signals in the brain, that is, the neural data of the brain’s analysis and decision-making process, and it can already distinguish vision, hearing, and language; the accuracy of its data is unknown. The nerve data of the limbs is extremely easy to measure. In other words, NAI applications can be carried out without collecting eye nerve data and ear nerve data.

[Solutions based on the social harm caused by my infringement]

1) Actively advocate that the police actively solve the case with existing laws. If the server is located overseas, I will actively support the Chinese police based on relevant global contributions (human brain neural artificial intelligence ethics research, NAI global fair development and governance plan, faith tolerance, new crown medical research).

2) Do not advocate the use of social radical actions for infringing neural technology, so as to ensure that the entire incident is under control and does not lead to radical actions by the technology infringer. At the same time, actively dialogue and encourage them to use methods that contribute to humanity and the Chinese people to obtain amnesty according to law after being controlled.
3) Social level: From the perspective of security and democracy, any social application of this technology is not encouraged; non-contact long-distance brain-computer interface (passive radar within 500Hz) is strictly restricted; all radio equipment within 1000Hz is strictly prohibited; any intentional creation of contradictions and intentional framing under the pretext of interpreting non-behavioral neural data is strictly prohibited; any behavior of interpreting neural data close to the person is strictly prohibited, and it is classified as brain control that causes the person’s brain to fail, and is fully responsible for the society
4) Economic level: It is estimated that NAI automation technology will be developed in accordance with the law; at the same time, from the perspective of fair development, economic methods will be used to solve potential and existing social problems that arise in the process of economic application of this technology, so as to facilitate continuous advancement, and it is advocated to adopt this concept for global
governance and development
5) Legislative level: Combined with relevant social practices, actively and orderly promote special legislation on relevant neural technology and artificial intelligence technology
6) Individuals should adopt methods that try not to contact society as much as possible, depending on the situation and social response, as well as personal objective basis