Suno

 I wrote the lyrics, the AI composed and sang, and the albums came out in minutes:

https://suno.com/@senron

Sora

https://openai.com/sora 

Will the next step be 3D video generation models as world simulators?

And then the next neural signals state generation models as brain simulators?

En la pasinteco, inteligenteco en maraj kaj ĉielaj medioj estis relative simpla.  

Nun, faru bone ĉe kompleksa inteligenteco en teraj kaj marfundaj medioj ankaŭ kontentigas la unuan.

Ĉio estas duobla traktado de kompleta logiko kaj probabla sperto, duflanka aliro.  Ne estas certe ke la kompleta logiko aŭ probablo sperto, kiun oni konsumas pli da kalkuloj.  Ĝi estas ofte konata post kurado de tempodaŭro, kaj ĝi eble ŝanĝiĝas denove.

立体环场空间状态扫描采主动式射频回波测距在军用上会暴露行踪、在民用上亦易造成干扰；有一种演算法可以从多角度被动式摄得的平面影像合成立体模型可是运算量极为庞大。道路上车辆颇多，有车联网共享，多角度影像是随时大量在取得，若各车备多些运算资源就可以分担中心的运算。车多时倾向多车组成的 near cloud 而疏解与中心的通讯负担，车少时附近运算资源不足而远端通讯带宽反就充足所以倾向 far cloud，如此即需要底层分布式操作系统自动调节运算分布方为完整的 Cloud。

What the neural networks

As the processors in computers were never called the electronic brain, the neural networks we usually called before should not be called neural networks. They belong to the 5th generation computers; the real neural networks belong to the 6th generation computers.

The imitating and synthesizing

The audio and video communication can be imitated and synthesized, several times the counterfeitings from the opponent were seen through in the reaction mode of the temporary interaction. Will what signals the AI imitates be not only the outer representation in the next? It will be much more and more difficult.

What is real?

As there will be the real Cloud, there will be the real neural network. But the real Cloud may come true in several years; the real neural network may be several 10 years.

🤖

La vizio de la Oksigena Plano je la komenco de la jarcento estis lasi homojn helpitajn per artefarita inteligenteco laboras aŭ studi ien ajn kaj iam ajn, sed hodiaŭ ĝi estis evoluigita por esti monitorita de la artefarita inteligenteco kiam homoj laboras aŭ studas. La originala celo liberigo fariĝis kontrolo.

The vision of the Oxygen Plan at the beginning of the century was to let people assisted with the artificial intelligence work or study anywhere anytime, but today it has been developed to become monitored by the artificial intelligence when people work or study. The original purpose of liberation has become control.

Virtual Reality 和 Reality 是不是同一回事？

在于 Spatial Mapping 的精准度与解析度，只要夠接近，基本上在 VR 里 run 的就是在 reality 实际 run 的。如外在世界不到脑里形成不失真的 VR，大脑又何从思考做出判断以传令予肢体应对 reality ？对 reality 有准确的认识，在脑里 VR 演练过多回，到 reality 实行时方能一次做对。在 VR 里失误不会有损失，在 reality 失误就是实质的损失。

智能行驶

Spatial mapping 是 Google car 首先带到民车的，之前是 DARPA 在 tank 上面搞，tank 才是在人车稀少的野地测试，Google car 是在公路上跑有几年时间 0 事故，那时我还很呕说我光是 VR 模拟就频频当机，他们是怎么办到的？智能车开始出现事故是这一两年的事，看到有一辆直直往分隔岛冲撞去，很惊讶那么明显的障碍物怎么可能没侦测到？得知是把分隔岛另一边道路的右边地标线当成是自己车道的左边地标线，这不是早在世纪初还没有 spatial mapping 以前就开始在公路测试的平面影像辨识与移动追踪吗？现在不是已经多了 spatial mapping 吗？就算看错了地标线怎么会不知道前方有障碍物？火场寻径与道路寻径是通用同套，管它是火场还是人车都是禁止穿越过的障碍物，这还不需用到类神经网络，这是不经过思考几近反射的简单立即判断，除非是同时出现多个多向快速移动中的障碍物才伤脑筋。在 VR game 里都 run 到驾轻就熟很平常的动作，在 reality 意外是怎么发生的？确定不是传感器故障或处理器当机？

AI 寒冬？明明现在我就热到要吹冷气降温

从上世纪末到这世纪初科技界所谓的 AI 寒冬是指 NN 克服不了技术瓶颈表现差强人意沈寂到了近年才在技术上取得突破，而今年非技术人员所谓再一次 AI 寒冬算是属于达康类型的商界情景。当前并非所有 AI 都用 NN 为佳如同人脑亦是有些事用精准逻辑表现较好、有些事用模糊感觉表现较好而采另一套反而表现较差，早就说过要双管齐下同时进行而何者临场结果可接受即采何者，所以不要单方面看到某些时候 NN表 现不如预期就开始在说什么泡沫寒冬。自动驾驶开始出现事故当然是要去改进而非因噎废食，试问在 AI 以前有因为交通事故频传就都禁车而不改进车体的机构安全吗？从来想不劳而获的都是搞金融的，泡沫就只有股市商式炒作在泡沫，技术是不能不持续改进，为的是实用。

Basic AI

No matter the Mathlab solves the formulas or the Prolog drives the rules or Siri or Cortana processes the natural language, they are functions processing the strings as parameters. What I want is the processor directly runs in formulas, rules, and sentences as programs but not only processes the string as text.

Patterns ↔ Senses

试想，若你的听觉受器是如视觉受器一般是阵列收讯，声讯对你来说是不是如同视讯画面一般？如此你会直觉两种收讯是不同的处里吗？是刚好你天生听觉受器就仅有两个单点，才会直觉视与听不同，在你进化前暂把声讯当成是1x1的2D画面串流，那直觉就会是一样的处理了。

All kinds of signals are patterns, the patterns are resolved to be the senses, and the senses are patterns too.

传统处理器在理解文字是一维循序串列处理，现在将每个字当做一个像素的值看待，不就是一张二维图像？那你觉得可不可以平行并列处理？整个理解过程是整面 operators（ 动词、介系词、连接词... ） 结合旁边的 operands（ 名词、冠词、形容词、副词... ）同时在发生，整个画面感觉就像粒子接触结合的化学反应一般组成了 Sense structure 以于脑内进行思维。

同理逆推，由 Sense structure 合成 pattern 输出是咋样的情形可想而知。

看似不同的事其实是同一件事

我做的是 Dynamic Shell 的 strategic control，是立体全空间的寻径，非仅一机动单位的寻径还有多单位彼此之间的协调寻径。当 shells 在外头跑完停靠聚落即磊堆宛如建物，由一堆 shells 为移动式房间组成的建物，不是把车艇停在一区然后步行到另一区的房间而是车艇即房、房即车艇。每个贝壳进出聚落磊堆的时间不一，即使在休息中也自动挪移以让欲出之单位轮流转至外边，所以停靠聚落一样需要协调寻径，于是无论在外跑或停靠聚落都是相同的控制。

若也用这些 Dynamic Shells 来做物流配送，聚落磊堆时不就相当物流中心的载货移动机台？所以外面的配送和里面的配货有何不同？完全相同嘛！Shells 既是移货机台也是活动仓储，货在 shells 间移交时又相当于什么？自动贩卖机。而自动贩卖机内部的调货和外部的调货不又是相同？差别只在于大小而已。

于是总结：

A structural object is composed of the structural objects the same as itself, and the structural objects compose to be a larger structural object the same as themselves. It is the infinite space recursive structure.

The natural language

As facing with people from all over the world, language or words are not useful. The best way to communicate is passing dynamic graphics: the bionic way is to make our skin cells color-variable as the screen pixels controlled by the nerve; the applied sci-tech way is to use the #HoloLens. Or we directly use the brain wave communication originally in nature or artificially assisted by the electromagnetic equipment.

Yeah! What the natural language we so called in the artificial intelligence in the past is not the natural language. It is the social language. The real natural language is the dynamic graphics language. The natural user interface offers it.

How the Cloud virtual reality is synchronized by the real world

The viewers watch the virtual reality from the Cloud directly but do not change it, they change the actors in the real world and then the sensors in the actors update the virtual reality in the Cloud. In the pure simulation mode, the actors with sensors are replaced by virtual ones in the virtual reality in the Cloud. 

錄一段語音取樣即可模仿一個人的聲音說話，錄一段手寫取樣即可模仿一個人的筆跡寫字，這是辨識技術的反向合成技術。所以，証據？証據是個屁？

指紋和DNA也都可以輕易栽贓，實在已不適合當絕對證據僅能參考。那怎麼辦？沒有任何一種認証是不能被假造，但要所有的跡證都被假造到天衣無縫也是難上加難，所以就是多重認証多重證據，多一項就是多一分正確性與可信度。

It is necessary to set the weapon on top of the helmet by the tranmission control to aviod exposing the head and the eyes outside the bunker, and the helmet itself is a drone too.

It is not a game. It is the simulation by the playing from people. Everything occurring in the real world is scanned into the virtual reality; everything happening in the virtual reality could change the real world too but not must be immediately. It could simulate until making sure everything is ok and there is no any problem then let the drones update the real world in according to the new right virtual reality.

The completely bionic artificial neural network

The neural signal scanner is harder to implement than the neural network scanner. The scanner is one, and the other one is the container. No matter how the RD progress of the scanner is, the container is easier to implement. We can use the current virtual reality to record the future scanned 3D image and then simulate the real work of the neural system with slower performance. However, if we want it to run faster, we must make the scanned neural network to be the hardware. In addition to that, the way to have virtual reality in the completely bionic artificial neural network is different from the current processor system, and it will run completely the same as the neural system in our bodies. 

The question is: even one day the scanning technology reaches the resolution of the neural ion stream, how to generalize the formula for simulating out the next state from a series of the scanned states and how to be sure it is working the same as the original brain? 

Record the entire brain map at as much spatiotemporal resolution as possible and for as long as possible. This is the most popular way to commemorate life in contemporary times, just store it in the cloud first. For a large number of continuous brain maps, perform a regression analysis to find out formulas or train the neural network to generate the next time point of the brain map, and then compare it with the original brain maps. This will go through a lengthy iterative process until the simulated brain is fairly consistent with the original brain map record. With the development of science and technology in the times, more people of later generations have more detailed records of spatiotemporal resolution, and simulation technology is also constantly improving. But this is a kind of clone, and the original will be still gone.