Why did you write the book "Questioning Artificial Intelligence-From Cambridge to Beijing"? Simply put, it started because of doubt...
Since returning to China from a study visit in Cambridge in October 2013, I have become more and more disturbed by the new wave of artificial intelligence represented by machine learning. It is not a complete denial of the role of machine learning. Enumerated), but worried about its obstacles and fetters to the future intelligent field. As Michael I. Jordan, a leading figure in machine learning, said: “I hate to call machine learning AI. The former is just a part of it. IA (Intelligent Enhancement) may be more suitable to define the development of intelligent technology today. ...People can't have much expectations for the so-called AI. These are just building a'smart' automation system." So how big is the gap between the actual capabilities of current artificial intelligence and the intelligence that people really expect? In what way will the future intelligence be presented in people's lives? Is the origin of artificial intelligence as written by many experts and scholars in speeches and academic papers? ...With these curiosities and questions, combined with some things that I saw, heard, read, learned, studied, thought, realized, checked, and found things that I saw, heard, read, learned, studied, thought, realized, found, and found outside, During this period, I got the help and inspiration of many mentors and helpful friends, and this little book was initially formed, and it also laid a foundation for the next little book. The following is the beginning part of my suspicion. I hope you will not hesitate to make corrections. Thank you for your insights and insights!
Many things and things often happen at the same time in the four domains of the physical domain, the information domain, the cognitive domain, and the social domain. Among them, there are factuality, value, and sometimes responsibility. In contrast, the factual data input, reasoning, and decision-making changes in the human-computer fusion are relatively small, while the valuable information/knowledge cognitive process is more flexible. How to clarify the synthesis mechanism of factuality and value will become The key issue of human-machine integration of intelligence is also the root cause of the "unintelligence" of artificial intelligence: "Something (something) is really priceless (value)".
The existing intelligence is basically a formal factual calculation, and lacks an intentional value calculation, so it only solves a very small part of the intelligence problem-part of the automation is far from the expected intelligence... …, this is because artificial intelligence or automation only solves the external input, processing and output, while ignoring the more important internal input, processing and output.
Input is divided into two types of input, internal and external, among which the eyes, ears, nose, tongue and body are external input, reflection and enlightenment are internal input, and language is a combination of internal and external input.
Processing is also divided into two forms of internal and external processing, of which bottom-up (axiom) calculation is external processing, top-down (irrational) calculation is internal processing, and calculation is a form of internal and external integration. .
Output is divided into two output forms, internal and external, among which logical fact-based decision-making is external decision-making, intuitive (non-logical) value-based decision-making is internal decision-making, and fact-value-shaped decision-making is an output form that fits internally and externally.
We can collectively refer to external input processing and output as factual formalization, and internal input processing and output collectively as value intentionalization. Then the essence of human-computer fusion intelligence is a community of factualized facts and values.
Formal facts calculate the approximate linear relationship of various parameter variables, and the value calculation of intentionality is not essentially dealing with the relationship of each parameter element, but dealing with the interaction of various relationships. Intentional value calculation is equivalent to re-arbitrary segmentation, reorganization, free arrangement and combination of various heterogeneous formal calculations.
Any intelligent system needs error correction. The concept of calculation, that is, computer calculation + human calculation, can be used as an error correction mechanism system mechanism design to realize Wittgenstein's non-family similarity compensation verification.
True intelligence or artificial intelligence is not what an abstract mathematical system can achieve. Mathematics is just a tool, and what it achieves is only function rather than ability. Only talents produce real abilities. Therefore, artificial intelligence is the product of the interaction of people, machines, environments, and systems. The future intelligent systems are also the insights derived from machine calculations combined with human calculations.
Gods exist, and they are common in good Indian movies!