- Published on
[Science of Systems] Humans and computers are essentially the same in the sense that they solve problems of satisfaction.
Roughly speaking
- China advances with Biden victory. When thinking about autonomy controlled by AI.
- Simon's reasoning led to the inherent absence of a distinction between humans and AI.
- Humans need to understand the emergent aspects and think about differentiating and collaboration with AI.
The beginning of autonomy through AI
In 2020, the US presidential election showed that pro-China Democratic Party Biden was elected, and it can be said that China's hegemony in diplomacy has almost established itself.
China has an advanced urban structure. The city is equipped with surveillance cameras and an automatic reporting system, and if you ignore a traffic light, for example, you will receive a notification of the fine to your home a few days later, or you will be encouraged to pay the fine via the app.
I previously wrote about unmanned governance in an article about Future Blue Ocean, but it has already been realized in China. Japan will likely begin a society of surveillance using AI.
Long before AI became popular, some economists have been analyzing systems and predicting the future since the 1960s. It's Herbert Simon.
Simon previously introduced the special lecture stories in a trilogy. In short, this is a smart person. A genius of reasoning. The Japanese translation of the special lecture is the best material I came across this year, and the "science of systems" we will be introducing today is also on the verge of being a lifetime best collection.
Herbert Simon How humans can make rational choices? (Reason in Human Affairs)
Simon Strange similarities between evolutionary theory and corporate pursuit of rationality
Simon Pursuing rationality in the social system and not being fooled by the media
This book derives how human society addresses the problems faced by the contrast between natural and artificial objects. The story goes all over the world, and to be honest it is extremely difficult to understand. However, if you read it over and over again without giving up, you will see that it is developing an incredible reasoning.
Among these, computers are one of the human decision support tools. Simon shows that the structure of this computer is the same as that of a human. It was indeed 1967.
The science of systems is diverse, so this article alone cannot show the content at all. However, I hope to continue to share some of them in an easy-to-understand manner on my blog.
There is no essential difference between AI and humans
So, is Simon saying that what is the same between AI and humans?
According to Simon, this can be derived from internal structures and from an approach to the problem of satisfaction.
Internal structure similarity
First, let's start with a simple comparison of the internal structure. According to Simon, modelling intelligence means that humans and computers are essentially the same.
First, let's model intelligence. Intelligence refers to the function of a symbolic system. So what is a symbol system? The features are listed below.
- A symbol system has a symbol pattern like a mark and creates components (expressions) of the symbol structure
It simplifies and allows you to think about the environment by modifying, modifying, duplicating, and destroying it.
<br/>
<br/>
- Symbols allow the environment to be modeled to some degree accurately and theorized.
<br/>
<br/>
- For symbol systems to be useful, windows and hands are required open to the outside. In other words, there is a need for a mechanism to obtain information from the external environment and convert it into internal symbols, and a means to inspire the environment.
<br/>
<br/>
- Symbols can indicate the very process that the symbol system interprets and executes. To indicate, programs that determine the behavior of the symbol system can be stored in the system as memory, along with other symbol structures, and can also be performed by inducing the memory.
<br/>
Thinking that intelligence has the above characteristics, we take a look at computers and humans again. Computers are made up of glass and metal, and human brains are made up of blood and flesh. Computers interpret digital signals, humans build symbol systems for mathematics and logic, and use intelligence with paper and pencils. In chronological terms, humans created computers, so it may be possible to introduce human symbol systems into computers, and the collaboration between computers and humans allowed symbol systems to actually interact with the environment. In any case, the internal structure is similar in terms of intelligence.
Similarities regarding approaches to satisfaction problems
Computers and humans are also similar in terms of problem definition and problem solving approaches.
Simon sees problems in the world as design problems, and says that this will result in problems of satisfaction.
The problem of doing the best in one's field of view regarding how to maximize utility, taking into account the constraints of the external and internal environments, for a set of possible worlds, is called the satisfaction problem.
Computers attempt to get the correct answer by deriving utility functions strictly for satisfying problems. Furthermore, by comparing the search that derives utility functions with the computational cost of the utility functions of alternative options with the interruption of search, the analysis of the satisfaction problem can be interrupted with some satisfaction conditions. Of course, supercomputers and machine learning can also overcome these limits by themselves.
Humans solve the problem of satisfaction through intuitive judgments produced by short-term and long-term memory, and pattern recognition that long-term memory has. Of course, he is also good at heuristic analysis, where information is collected during searching and solving problems. It can be thought that people who are incredible in chess and shogi are trained to intuitively judge the pattern recognition, and that heuristic analysis is the way in which corporate activities are resolved through higoro meetings and other events in modern society.
According to Simon, the mechanisms for the satisfaction problem are summarized as follows:
Computers or humans, or computer-human collaborative complexes (like AI and Ghost in the Shell) are importing and exporting various ideas from one intellectual domain to another, as to how computer-human collaborative complexes (like AI and Ghost in the Shell) solve problems and achieve their goals in a significantly more complex external environment.
The raison d'etre of human existence in the AI era
From here on, I will present my personal views on a society where unmanned governance is progressing.
First, I believe that the transition to unmanned governance is inevitable given the cost merits, and that sooner or later will arrive in which humans lead cramped lives in surveillance society.
This is because, essentially, if humans and AI solve the same problem, there are a lot of routine problems that will give AI the chance to win. It is inevitable that existing jobs will gradually disappear.
The most advanced AI technology is military technology. Have you seen footage of the current conflict between Azerbaijan and Armenia? Suddenly, a drone plunges into an Armenian truck and self-destructs. This drone is an autonomous AI and has the ability to chase soldiers and self-destruct. The current concept of soldiers will gradually disappear, and war will take place unmanned.
If we strengthen the surveillance society, violations will be automatically cracked down. Simon has kept a keyword in mind when it comes to where human food is being supported. It's collaboration.
Computers are good at solving existing problems, but humans are good at omitting calculations using intuitive judgment. Problems that address new problems creatively remain as human expertise.
There is still a good chance that an ethical society will be created so that computers can be used in a way that respects humanity.
That is why many people need to read the science of systems, understand the similarities and differences between computers and humans, develop ethical values, and handle computers correctly.
In particular, people who know the power of computers need to be altruistic and handle computers with caution so that they do not lead to human destruction.