In the 50s of the last century, not long after the end of World War II, people began to study and design intelligent systems. As a branch of informatics, mankind began the earliest research on artificial intelligence. In the '60s, there was a lot of confidence in the development of computers, and people asserted that "in 20 years machines will be able to do what anyone can do......The problem of creating AI within a generation will be substantively solved".
However, soon, the development of artificial intelligence entered a period of stagnation in the 70s and 80s, and it was not until the 21st century that the research of artificial intelligence entered people's field of vision again due to the development of machine learning and data analysis technology. Recently, GPT has reignited the craze for the vision of this technology. The predictions of the '60s didn't come true, and the development of artificial intelligence was not as rapid and optimistic as one might think. The most fundamental reason for this is the lack of understanding of the limitations of computational theory. To build a machine that behaves like a human, we had to understand how "human intelligence" worked, and neuroscience wasn't that powerful at the time. The articles in this series are what I read recentlyUnderstanding and Changing the WorldA summary of the takeaways from the book. Writing down words is also a process of organizing your own thoughts. (I highly recommend everyone to take a look at this book).
This article will discuss"The Divide Between Human Intelligence and Machine Intelligence".and"The difference between an automated system and an autonomous system".If you can read these two topics, then there will be more new understandings of artificial intelligence beyond technology and under the surface. In today's article, I will mainly introduce the main manifestations of the human mind and the gap between machine computing and machine computing. Think fast and think slowlyIn the book Thinking Fast and Slow, author Daniel Kahneman details the two ways in which the human mind works:Think fast and think slowly
"Slow thinking" is the thinking that reacts slowly but processes consciously, such as mathematical calculations and logical reasoning, is similar to the computational model of computer programming, which is far better at dealing with such problems than the human mind. Another kind"Quick thinking" is thinking that reacts quickly and automaticallyFor example, the coordination of the body's limbs when walking, eating, and the body's alertness in the face of unsafe environments, etc., the brain completes this series of conscious behaviors almost instantaneously, and only consumes very little energy. Today's artificial neural network computational models are designed to mimic the fast thinking of humans. These two ways complement each other and form a human way of thinking. Since its birth, the slow thinking system has gradually created a series of processes, which are then automated by the fast system. For example, babies work very hard to call "mommy" when they first learn to speak, and with practice, the act of "calling mommy" becomes more and more proficient and automated. The same goes for learning to ride a bike and drive. So in a way, this is the most fundamental manifestation of human intelligence. 
From computers to human intelligence
Are today's computers intelligent? Many people think that "intelligence" is manifested in the formation of decisions, the solution of complex, definite, well-defined problems, which obviously does not reach the height of human intelligence. The important characteristics of human intelligence are "autonomous behavior" and "adaptation to changes in the internal and external environment". This is the key to the human brain's ability to create new knowledge, understand situations that have never been encountered before, and set new goals. But unfortunately, current machine learning doesn't do that, and we're seeing more so far"Programmer Intelligence"., that is, to execute the commands that the programmer describes in advance with symbols. 
Only when the day comes when computers can perform a large number of tasks autonomously without human intervention and are able to adapt to changing internal and external environments can we say that the gap between artificial intelligence and human intelligence is disappearing. In order to understand human intelligence, we need several crucial concepts (which can also be understood as the behavior of the brain):Awareness, understanding, common senseAbout consciousness
Regarding consciousness, you can understand it through this sentence:The brain's ability to "see" how it acts in the semantic models of the external and internal worlds。This reminds me of the domineering (**ability) of One Piece. Consciousness refers to the state of human subjective experience and perception, which is our perception and understanding of ourselves and the external world. 
Consciousness has several characteristics such as:Subjectivity(Only individuals can experience it).Initiative(able to actively choose and decide to a certain extent),Continuity(Mobility of Consciousness) andSelective(Ability to select the object of attention). The counterpart to consciousness is the non-conscious, non-conscious mental processes and information processing, namely the fast thinking mentioned above. The human brain continuously produces so-called "common sense" through the continuous "understanding" of the objective world. It's about understanding
By "understanding", we are able to:Relate what is observed to relationships that are already grasped in the mind。For example, if we see something falling to the ground at an accelerated rate, we understand that this is because the Earth's gravitational pull is at work. The world is complex, and in order to make it easier to understand the various phenomena in the world, the human brain tends to distinguish between the main and secondary factors that affect the phenomena. However, sometimes phenomena are too complex, such as climatic, economic, and social issues, and this overload of complexity leads to limitations in human understanding. This limitation of understanding is further limitedThe construction of theories(Einstein did not unify the theory of relativity and quantum mechanics until his death) andThe complexity of workpiece manufacturing(Due to the lack of basic biology, it is impossible to find an effective method for cancer in the short term). About common sense
Much of human intelligence falls under "common sense". The emergence of common sense means that a certain type of thinking in the brain has changed from full thinking to fast thinking. The brain uses its semantic model to evaluate what is happening in the environment and what is possible, and the semantic model that is constantly accumulating experience enriches itself every day. In order for a computer to behave in a human, we must give it a semantic model. Theoretically, if we can analyze and formalize natural language, build a semantic network of relationships between concepts according to the hierarchy, and add rules for representing and updating knowledge, we can build such a model. 
For example, when defining "father and son", we need to imagine all the relationships and rules that the word contains and formalize them. However, in the last 50 years, we have not made substantial progress in this direction. By understanding the characteristics of human intelligence, we can better understand what it takes for a machine to implement a similar intelligent system. In the next article I would like to talk about it"Automatic system".and"Autonomous systems".The implementation of autonomous systems will mean that the boundaries between the two types of intelligence are becoming increasingly blurred.