Summary: Explain 5 questions that need to be answered in formulating artificial intelligence ethics

This article is based on the author of Internet evolution and computer doctor Liu Feng, a computer science and technology strategy consulting institute of the Chinese Academy of Sciences and Tencent Institute jointly held the speech content of the "2017 Artificial Intelligence: Technology, Ethics and Law Seminar" in Beijing on April 13. . The topic of the speech is "The relationship between the formulation of artificial intelligence ethical rules and the direction of biological evolution"

This article elaborates in detail the five questions that need to be answered in formulating artificial intelligence ethics, and proposes whether a consensus can be reached on whether biological evolution has a direction. It will be the key to whether the formulation of artificial intelligence (AI) ethics can form norms in the future. The research report compiled based on the speech "2017 Artificial Intelligence: Technology, Ethics and Law Seminar" is as follows:

In layman's terms, ethics refers to "the relationship between man and man and nature and the rules for dealing with these relationships". In the thousands of years of civilization history of mankind, we still do not have a unified, standard and clear ethical system, only There are some general principles that most people recognize. The ethical impact brought by artificial intelligence has become more prominent due to the imperfection and controversy of ethical issues.

The Titanic (RMS, also known as Titani), is an Olympic-class cruise ship that sank after hitting an iceberg during its maiden voyage in April 1912. The Titanic shipwreck was one of the worst shipwrecks in peacetime. More than 1,500 people on board were killed. Which group gains priority for life during the escape process? It has become an important problem faced in the escape process. David Popno, a famous sociologist at New Jersey State University in his book "Sociology" wrote: "... Unfortunately, the lifeboat is not enough. Although many people (more than 1,500) were killed, passengers were careful to abide by the social norms of 'priority to rescue women and children', which made the British public and government face this huge disaster, and "some comfort can be found" - statistics show that "69% of the passengers survived, while only 17% of the male passengers survived."

2012 is a disaster film about global destruction, released in the United States on November 13, 2009. The movie tells the story of the protagonist and people from all over the world struggling to survive. In the face of disaster, all kinds of things appear in the world. In the movie, when Air Force One can only accommodate one person on board, the US president lets the physicist leave safely, and when he stays, he says that "a scientist is more important than dozens of officials."

The train is driving rapidly and cannot stop quickly. There is a fork rail directly in front, which is still in use on the left and is deactivated on the right (but the train is still safe to drive up). There were 5 children playing on the tracks that were still in use, and one child playing on the deactivated tracks. The train arrived at this time. Following the original route of the train, it will drive to the left and run into 5 children. If you drive to the right, you will only run into 1 child. If you were a switchman, would you turn the train to the right?

When Boston Powered Robot was moved, it was attacked by test scientists, causing it to stand unsteadily and fall. Internet video viewers protested, believing that the robot's rights were violated.

"Robots must not harm humans; robots must obey human commands unless this command contradicts the first principle; robots must protect themselves unless this protection contradicts the above two principles." The above are famous science fiction writers, " The "Three Principles of Robots" proposed by Asimov, the father of robotics, has been cited by countless science fiction novels and film and television works in later generations and is regarded as a classic.

In Asimov's eyes, robots are just cold machines and tools to serve humans. Since it is a tool, there is naturally no need for any rights. Therefore, when "protecting oneself" contradicts "not harming humans" and "must obey human orders", the most basic right to survival of a robot can be sacrificed without hesitation. .

However, in recent years, in the context of the rapid development of robot technology and the increasingly close relationship with humans, discussions on robot rights have attracted increasing attention, and scholars and the public have participated.

5 issues that cannot be avoided when formulating AI ethics

1. Who should formulate AI ethics

2. Develop AI ethics from whose standpoint

3. What principles should be followed to formulate AI ethics

4. Who is AI ethics for reference?

5. Can AI be regarded as a living body with the same authority as humans

In the mid-19th century, Darwin founded the scientific theory of biological evolution, and Darwin's theory of evolution with natural selection as the core is one of the greatest theories in the history of human science. Unify all disciplines of biology. The influence of Darwin's theory of evolution is not only limited to the field of biology, but also provides a new worldview, life view, cosmology and methodology, which has an impact on almost all scientific and humanities fields. Ethical issues including artificial intelligence will eventually be influenced by Darwin's theory.

Darwin's theory of evolution excludes teleology (which was praised by Mayer as one of his great contributions), and Darwin argues that variation is random, using the word "nature" that is not inferior to God to explain its carefully constructed evolution. The core of the theory. However, Darwin also had his confusion and helplessness. He lamented that "this vast and incredible universe... is actually a blind opportunity or a product of inevitable feeling that it is very difficult or even incomprehensible" (2-Xie Ping. 2014. The Origin of Life ——The abandonment and innovation of evolutionary theory.)

Darwin's theory of evolution believes that biological evolution is not an evolution from low to high, and humans are not higher than other organisms. Darwin imagined the biological evolution process as a large tree that constantly grows and branches, and all existing organisms are located in one of the trees. At the top of the small branch, it is difficult to say which one is more advanced, and it is meaningless to compare the high and low levels between the species that exist simultaneously. There is no predetermined direction for evolution. The evolution tree does not have a trunk with humans at the top. Humans are just an ordinary branch on the evolution tree. From the perspective of Darwin's theory of evolution, humans are not more advanced than mice and ants, and humans cannot represent the direction of development of organisms.

Darwin's theory of evolutionary view on the lack of direction in life will have new progress and breakthroughs under the new technological progress. Let’s discuss this issue from the research progress of brain science, the Internet and artificial intelligence. Propose that the progress of brain science, the Internet and artificial intelligence will jointly demonstrate the direction and goals of life evolution from different angles.

Although the expressions of organisms vary greatly, the core of organisms, the brain does show obvious directionality. From single cells to humans, the brain is becoming more and more complex and has a higher degree of intelligence. After reaching the highest point of human beings, new forms of evolution emerged from organisms, and humans joined forces to evolve together through the Internet.

Nobel Prize winner Australian scientist John C. mentioned in his book "The Evolution of the Brain" that "the brain of a living creature evolves from the brain of a fish to the brain of a reptile, then to the brain of a mammal, and finally to the brain of a The human brain. If we dissect the human brain, we can clearly see that the structures of fish, reptiles, and mammals are clearly distinguished in the human brain. ".

Conclusion of Internet Evolution Theory "The Internet will evolve in a direction that is highly similar to the human brain. It will have its own visual, auditory, touch, and motor nervous system, and its own memory nervous system, central nervous system, and autonomic nervous system.

On the other hand, the human brain has evolved all the Internet functions at least tens of thousands of years ago, and the ever-evolving Internet will help neuroscientists uncover the secrets of the brain. Scientific experiments will prove that the brain also has the same search engine, the same SNS system, the same IPv4 address coding system, and the same Cisco routing system. . . ”

The Internet of Things is the germination of the sensory nervous system of the Internet brain, and cloud computing is the germination of the central nervous system of the Internet brain. Industry 4.0, Industrial Internet, drones, intelligent driving, 3D is essentially the development and germination of the Internet motor nervous system, big data It is the basis of Internet brain information, and AI is the basis of the generation of Internet intelligence, wisdom and consciousness.

When organisms evolve to this level of humans, humans will unite and evolve together through the Internet. And the result of this co-evolution is. The Internet connected to humans is highly similar in structure to the brain step by step. It continues to spread with the expansion of humans in space. If there is enough time, it can be foreseen that after a sufficient time point, the space, the brain, and the Internet will be Those will be integrated into one and evolve into a mature Internet-like brain architecture.

The conclusion of Darwin's theory of evolution is deduced forward, that is, organisms have common ancestors, and through natural selection, the species of humans emerged. The conclusion of the Internet evolution theory is deduced backward, believing that people evolve towards the Internet-like brain architecture through the technologies and tools they create. The element of "human" connects the evolution of the Internet with the evolution of biological organisms.

The criteria for determining standard intelligent systems: If any system (including artificial intelligence systems, human life systems) meet the following characteristics, it can be considered that this system belongs to a standard intelligent system.

Feature 1: The ability to obtain data, information and knowledge from the outside world through sound, images, text, etc. (including but not limited to these three methods).

Feature 2: Able to convert data, information and knowledge obtained from the outside world into knowledge mastered by the system.

Feature 3: Ability to innovate by using the knowledge you have mastered according to the needs generated by external data, information and knowledge. These abilities include but are not limited to association, creation, guessing, discovery rules, etc. The result of using this abilities can form new knowledge you have mastered.

Feature 4: It is possible to feed back the data, information and knowledge generated by the system to the outside world through sound, images, text, etc. (including but not limited to these three methods).

The establishment of standard intelligent models refers to the von Neumann architecture. The von Neumann architecture consists of five parts: calculator, logic control device, memory, input system and output system. By comparing the differences between the standard intelligent model and the von Neumann architecture, we can find that the von Neumann architecture can supplement two parts. Through this supplement, we can express humans, machines and artificial intelligence systems in a clearer way.

The first supplement is the innovative creation function, that is, the ability to discover new knowledge elements and new rules based on existing knowledge, so that it can enter memory for use by computers and controllers, and to carry out knowledge with external use through input/output systems Interaction. The second addition is external knowledge bases or cloud storage that can perform knowledge sharing, while external storage in von Neumann architecture serves only a single system. Therefore, expanding the von Neumann architecture can form a new architecture

Direction of evolution - the level division of intelligent systems

The basic characteristics of the level 0 system of intelligent systems exist in theory, but such artificial intelligence systems do not exist in reality.

The first level system of intelligent systems is a system and object that cannot interact with human testers.

The second-level system of intelligent systems, its basic feature is that it can interact with human testers, and the existence of controllers and memory.

The basic feature of the third-level system of the intelligent system is that in addition to having the characteristics of the second-level system, the programs or data contained in its controllers, memory can be upgraded or added without networking.

The basic characteristics of the fourth-level system of the intelligent system, in addition to the characteristics of the third-level system, are the most important that information and knowledge can be shared with other intelligent systems through the network.

The most basic feature of the fifth-level system of intelligent systems is that it can innovate and create, identify and identify the value of innovation and creation to intelligent systems, and apply the results generated by innovation and creation in the development of intelligent systems.

The most basic feature of the sixth level system of intelligent systems is that as time goes forward and tends to infinite points, it constantly innovates and creates the input and output capabilities of intelligent systems that generate new knowledge, and the ability to master and apply knowledge will also be approaching infinity,

Inspiring conclusions obtained from brain science, evolution of the Internet and classification of intelligent systems hierarchy

The evolution of the brain, the evolution of the Internet, and the division of intellectual hierarchies of intelligent systems all show obvious direction. For these three fields, their common point is that they carry the knowledge and wisdom of life, and are constantly evolving and improving.

The improvement and evolution of knowledge and wisdom are the core of biological evolution, and the direction of biological evolution and the level of biological evolution from the perspective of knowledge and wisdom.

The expansion rate of biological population knowledge base is the focus of biological evolution. The knowledge bases of other organisms have stagnated and are heading to a dead end. Therefore, there has been no further change in tens of millions of years and are in an increasingly lower position in the life circle of the earth.

In the past 10,000 years, human beings have continuously expanded and accelerated their knowledge and wisdom, and have made further huge leaps due to the invention of the Internet and artificial intelligence, thus gaining the dominant position of the earth's natural competition.

Direction and objectives of biological evolution

The direction of biological evolution is to continuously expand the knowledge base of populations and the ability to transform nature, and the goal is to achieve omniscient and omnipotent nature (circle of life, universe). The level of a creature is divided according to the knowledge base capacity it has and the ability to transform the world.

The impact and answers of evolutionary direction on AI ethical issues

1. Who should formulate AI ethics

Formulated by the most knowledgeable and most intelligent biological population

2. Develop AI ethics from whose standpoint

Standing on the standpoint of biological populations with the richest knowledge and the strongest wisdom

3. What principles should be followed to formulate AI ethics

Develop AI ethics in accordance with the principle that is beneficial to the knowledge and wisdom of the population that expands (where the rule-making person is located)

4. AI ethics is first of all for reference and use

For managers (government, entrepreneurs), researchers (scientists), developers (programmers and product managers) of populations (references)

5. Can AI be regarded as a living body with the same authority as humans

AI cannot be regarded as a living being with the same authority as humans. It shares some of the human knowledge and intelligence functions, but it cannot be replaced in terms of the most important aspects of creativity and review of creative value. More importantly, AI cannot determine its own evolutionary direction. And the evolutionary goals, there is no driving force for correct evolution. Its evolutionary power comes from humans and is still a tool for humans.

The Titanic lets women and children leave because they will be more helpful to the future of mankind and can better expand and inherit human knowledge and wisdom. In the 2012 movie, the president chooses to die and let young geophysical scientists survive. Because scientists have more knowledge and wisdom to contribute to solving human crises. In the dilemma of the switch, we need to analyze which side may make a great contribution to the future knowledge and wisdom of mankind.

But there is still a need to balance population interests and personal interests. How the ethical setting balances between population and individuals is also related to the urgency of the scene. For example, in an extreme environment like human extinction in the 2012 movie, and in ordinary scenes with low levels of crisis, group interests and individual interests. The allocation should be different.

It is necessary to form a consensus on the direction and goals of biological evolution, and then formulate the ethical rules of artificial intelligence (AI) to have a foundation.

If there is no consensus on whether there is a direction for biological evolution, there will be many difficulties in formulating ethical rules of artificial intelligence (AI). For example, humans and animals are at the same level. How to deal with the relationship between humans and animals when formulating rules. There is no direction for evolution, and artificial intelligence (AI) is formulated. Ethical rules should be given priority to protect who will have problems. If there is no determination of evolutionary driving force and evolutionary goals, the relationship between artificial intelligence systems and humans cannot be handled.

At the same time, the nature of emotions and consciousness is also related to the direction and goals of evolution. For example, happiness, excitement, and happiness are often related to the reactions caused by organisms to adapt to evolutionary trends. Sadness, anger, and disappointment are also often related to the reactions caused by counter-evolutionary trends. If not, The evolutionary driving force and evolutionary goals are determined, and it is impossible to judge whether the emotions of the artificial intelligence system are real emotions.

Important issues that need to be discussed in depth

The evolutionary driving force of AI is scientists, entrepreneurs, programmers, and product managers.

What is the evolutionary driving force for humans and organisms to have direction and goals?

Can the evolutionary momentum of biological organisms be applied to the development of AI? If successful, the AI ​​threat theory will have a basis for realization.

The setting of AI ethics will undergo major changes. If it is theoretically proven that it cannot, AI will never be able to surpass humans in all respects.

Appendix 1: Definition of Knowledge and Wisdom

Data is a conventional keyword to abstractly represent the number, attributes, location and their relationships of objective things, and save, transmit and process them in an artificial or natural way suitable for use in this field.

Information is a time-sensitive and meaningful data stream that is logical, processed and valuable for decision-making.

Knowledge is valuable information that is evaluated and reviewed, then deposited into the human knowledge base and structured.

Wisdom is the ability of human beings to analyze, compare and deduce the solutions based on the problems arising from the movement of the material world based on existing knowledge. The result of this ability to use is to mine out valuable parts of the information and make it part of the existing knowledge architecture.

Appendix 2: About the classification of intelligent systems

For the 0th level system of artificial intelligence system, its basic characteristics exist in theory, but such artificial intelligence systems do not exist in reality. In the hierarchical rules extended by the extended von Neumann architecture, some combinations can be made, such as information input, but information output, but information output; or information output, but information input; or innovation can be created, but knowledge base cannot increase. For these cases where corresponding system examples cannot be found in reality, we will uniformly classify them as "level 0 system of artificial intelligence system", which can also be called "specific system of artificial intelligence system".

For the first-level system of artificial intelligence systems, its basic feature is that it cannot interact with human testers. For example, there is a thought called animism that all things in the world have souls or natural spirits, and a tree and a stone have the same value and rights as humans. Of course, this view can only be regarded as conjecture or philosophical thinking from a scientific perspective. Judging from the hierarchical rule of "can you interact with the tester (human)" because objects such as stones cannot interact with humans, perhaps they have a knowledge base within them, can innovate knowledge, or can interact with other stones. But for human testers, it is a black box and cannot be understood by people. Therefore, objects and systems that cannot interact with the tester (human) can be defined as "level 1 system of artificial intelligence system". Examples that meet the level 1 classification include stones, sticks, iron blocks, water droplets, etc. that cannot be combined with An object or system in which humans interact information.

For the second-level system of artificial intelligence systems, its basic feature is that it can interact with human testers, and there are controllers and memory, that is, systems described by von Neumann architecture. Therefore, many household appliances are called smart home appliances, such as Smart refrigerators, smart TVs, smart microwaves and smart sweepers. Most of these systems have a feature, that is, although they have more or less control program information inside, once they leave the factory, they cannot update their control programs, cannot upgrade, and will not automatically learn or generate new knowledge. For example, in a smart washing machine, the washing machine will start the function of the keys that people press. Its function will not change from purchase to damage (except for failure). This system is able to interact with human testers and users in information, conforms to the characteristics described by the von Neumann architecture, and its control program or knowledge base has no longer changed since its birth. This system can be defined as "Level 2 system of artificial intelligence systems", examples include common daily sweeping robots, old-fashioned household refrigerators, air conditioners, washing machines, etc.

For the third-level system of an artificial intelligence system, its basic feature is that, in addition to having the characteristics of a second-level system, the programs or data contained in its controllers, memory may be upgraded or added without networking. For example, home computers and mobile phones are our commonly used smart devices, and their operating systems can often be upgraded regularly. For example, the computer's operating system can be upgraded from .0 to .0, and the mobile phone's operating system can be upgraded from .0 to .0, and the internal applications of these devices can also be continuously updated and upgraded according to different needs. In this way, the functions of home computers, mobile phones and other devices will become more and more powerful, and there will be more and more scenarios to deal with. In addition to home computers, many household appliances and robots have also begun to have interfaces, and the system can be upgraded through external devices. This type of system is obviously more adaptable than the second-level intelligent system. This system can interact with human testers and users information, but cannot interact with other systems through the "cloud". Its control program or knowledge base can only accept external devices such as USB and CD to perform programs or information upgrades. , can be defined as "level 3 system of artificial intelligence system", and examples include smartphones, home computers, stand-alone version of office software, etc.

For the fourth-level system of artificial intelligence system, in addition to the characteristics of the third-level system, the most important thing is that information and knowledge can be shared with other intelligent systems through the network. In 2011, the EU funded a project called "A project" aimed at allowing robots to share knowledge through the Internet. Helping robots learn and share knowledge can not only reduce costs, but also help robots improve their self-study and adaptability, and promote their popularization faster and larger scale. These capabilities of cloud robots improve their adaptability to complex environments. In addition to having the functions of a level 3 system, this type of system also has an important function, that is, information can be shared through the cloud. Therefore, this system can interact with human testers and users, and can perform information through the "cloud" Interaction, perform program or information upgrade. But all the information in this type of system is obtained directly from the outside, and it cannot generate new knowledge independently and creatively within it. This system can be defined as "level 4 system of artificial intelligence system", including Google Brain, Baidu Brain, cloud robot, B/S (/, browser/server) architecture websites, etc.

For the fifth-level system of artificial intelligence systems, the most basic feature is that it can innovate and create, identify and identify the value of innovation and creation to mankind, and apply the results produced by innovation and creation to the development process of mankind. When we expanded the von Neumann architecture, we added an innovative knowledge module to the original von Neumann architecture, which is to try to incorporate people into the extended artificial intelligence system concept. Humans can be regarded as constructed by nature. Special "artificial intelligence system". Unlike the first four levels, the biggest feature of humans and other living organisms is that they can constantly innovate and create, such as discovering gravity, periodic table of elements, writing new novels, creating new music, paintings, etc., and then through articles, letters, telegrams , and even the Internet for communication and sharing. Continuously conducting innovation and creation, and being able to identify the use of innovation and creation to oneself, this allows mankind to occupy the intellectual commanding heights in the earth's ecological environment. Therefore, this system can interact with human tester users, innovate and create new knowledge, and can interact information through "clouds" such as articles, letters, telegrams and even the Internet. This system can be defined as " Level 5 system of artificial intelligence system”. Humans are the most prominent example of Level 5 artificial intelligence systems.

For the sixth level system of artificial intelligence system, the most basic feature is that as time goes forward and tends toward infinite points, it constantly innovates and creates the input and output capabilities of intelligent systems that generate new knowledge, and the ability to master and apply knowledge as they move forward and move towards infinite points. It will also approach infinity. According to Christianity's definition of God, it can be seen that intelligent systems, represented by humans, will eventually be in sufficient time, under the circumstances of continuous innovation and creation and continuous accumulation of knowledge. From this perspective, whether it is the "God" in Eastern culture or the "God" concept in Western culture, from the perspective of the development of intelligent systems, it can be regarded as an intelligent system (including humans) The evolutionary state of future time points.