Rethinking on some problems of artificial intelligence

1, some people think: artificial intelligence is human beings in understanding themselves, know themselves. In fact, artificial intelligence is just human trying to understand themselves, because "Who am I" this coordinate origin is far from being determined ...

2, "Who am I" problem is the initial problem of autonomy, is also the coordinate origin of all intelligent coordinate system frame, memory is the direction of the vector (intentionality) in this coordinate system, and the storage of von Neumann computer system is different, the program rules and data information is not static, Instead of adapting to the human-machine environment system (so the individual brain meaning is small), the flexibility of this change often reflects the size of autonomy.

3, so far, the machine's storage is still formalized implementation, and the human is often visualized realization, artificial intelligence calculation is formalized the reality (some people say: The recursive function is Turing calculation, the secret of artificial intelligence. , and human calculation is often the result of the combination of objective logic and subjective intuition. The calculated forecast does not affect the result, the calculation of the expectations often change the future, in a sense, the situation is not calculated to perceive, but cognitive, self-advantages and disadvantages, is contrary to righteousness, is from the inside out of the attempt to correct, is the experience of verification ~ experience of the analogy migration.

4, language communication is the model of autonomy, is based on the interactive scene (not the scene), no matter how to test, are scripts and non-script response, the accuracy of the size can be judged by the human machine is not ... Some people divide the language into three fingers, that is, the naming, the heart, the point of three, and pointed out that the study of these three and their relationship has been the problem and challenge of artificial intelligence. Coincidentally, in the 19th century, the British scholars put forward the concept of being able to refer to, and think carefully, these are not only involved in the nature of things (can mean, feel) and the relationship between (refers to, perception) problem it. In fact, a word, a word, a piece of text can not be separated from the context of self-qualification, we know to (means) far more than we can say (can refer to) much more. If you don't believe me? Think of the people you see who can talk. Tracing roots, the reason is generally due to the reason of the transformation mechanism: sensibility is a rational wormhole, through the constraints of rationality and restraint, rationality is the perceptual black hole, restricting the emotional caprice and arbitrariness. It is the self-consciousness that drives the reason, and is enslaved by reason ...

5, the essence of intelligence lies in independent and "similar" judgment, is appropriate to grasp the "similarity of the benchmark" sense of proportion. One of the advantages of a man than a machine is that it can find patterns of things earlier in less data. One of the reasons for this is that the machine does not have the origin of the coordinates, that is, who is the "Me" problem. For people, things are non-existent---their existence is not objective, but we observe the results with subjective purposes, and the subjective and objective mixture is often the context of the product, such as the surrounding is (Being), should (should), to (want), can (CAN), The construction and deconstruction of the process of changing (change) are often carried out at the same time. In addition, even the same feeling (such as visual) has a specific point and abstract implication, the handshake at the same time, in addition to physical contact can be accompanied by psychological cues.

6, the human brain in the autonomous activities can produce "from the Euclidean space to the topological space mapping", that is, in the choice and control, the person can be based on the specific purpose of the difference, its basis for the similarity of the benchmark (not the European space on the proximity, but the sense of the contact network) is changing, And according to this decision to carry out the situation classification implementation.

7, a problem without boundary, unconditional, unconstrained solution is a philosophical study, the same problem has a boundary, conditional, constrained solution is a mathematical discussion.

8. The relationship framework theory of American psychologist Hayes (Steven Hayes): Our brains have a characteristic that it can generate new relationships between 22 related three things, in which language plays an important role, such as A>b, B>c, we will launch a>c. So, in the midst of complicated situations, does this relationship framework exist?

9. The free-regulated environment system triggers the inverse movement of the autonomous system, thus forming the multi-directional motion or multiple movement between the human machine and the environment, which leads to the contradiction and conflict.

10, fiction How to correct the real, how real feedback fiction. This is going to be a very tasty question.

11, human learning and machine learning is different: human learning is fragmented + integrity mixed, so the self-adaptability is strong, has been in the context of insufficient information (resources ~ such as space-time) in the situation of stable prediction and instability control, loss of pre-and out of control scene happens, so how two times, three times ... Many times in a timely manner of fast and slow multi-level feedback adjustment is more and more necessary, in this respect, people in the non-structural non-standard situation of the processing mechanism is better than the machine, and in the structured standardization scenario, the machine is relatively better than people. And this self-adaptability is cumulative, slowly will form a personalized reasonable expectations, so far, the independent (expectation + predictive + control) mechanism began to produce, and grow up ...

12, "the true symbol of intelligence is not knowledge, but imagination." Einstein said: "Imagination is more important than knowledge, because knowledge is limited, and imagination sums up everything in the world, pushes forward progress, and is the source of knowledge evolution." "Imagination is fiction, so fiction may be the essential representation of intelligence, déjà vu, specious, seemingly non-existent, but can be strongly tacit."

13, People's learning is causal relationship, correlation and even the integration of customs, these can be programmed, a lot of the present is difficult to describe clearly (such as some subjective feelings, Meurian knowledge, etc.), and machine learning dominant knowledge connotation efficiency far greater than the implicit concept of extension.

14, the relationship between rules and probabilities is a form of the cohesion, the rule is the existence of large probability, the nature of probability is not the formation of rules of the state. Habit is the unconscious behavior of rules, learning is the cumulative process of probability, including familiar analogy and unfamiliar correction part, in general, the former is unconscious, the latter is conscious, is a compound process.

15, the process of human processing of information is variable speed, sometimes the spontaneous release of the subconscious, sometimes semi-automatic conscious and unconscious balance, sometimes pure artificial slow, but this process is not pure information transmission, but also how to construct in the knowledge vector space to organize the corresponding grammatical state, and reconstruct a variety of semantic systems.