This chapter tries to give an appropriate perspective on artificial intelligence (AI) and review the work we have done and achieved. We list the achievements in the field of artificial intelligence over half a century and discuss the recent IBM Watson-Jeopardy Challenge. We also weighed the prospects of AI that had never reached the human level.
David Ferrucci
17.0 Introduction
First of all, we reviewed the importance of search, knowledge representation and learning in the construction of artificial intelligence systems, and gave examples to illustrate that appropriate knowledge representation helps solve the problem.
Secondly, we introduce a recurring theme in myth and literature—the attempt to create life or agents always encounter terrible consequences. Perhaps, we should give some warnings to the AI community.
This book illustrates the concept of problems that cannot be solved in computer science, that is, there is no problem of solving algorithm . We ask ourselves whether we can create human-level artificial intelligence, and that's the question.
Next, we review our achievements in the field of artificial intelligence over half a century.
Then, we discuss the Watson system of IBM. In March 2011, in a crowded TV match, the IBM computer defeated two of the most popular Jeopardy champions in the Danger Edge Challenge. Finally, we review several theories about creating life and explain intelligence and consciousness.
17.1 Blook out the main points—Overview
In Chapter 1, we began the journey of artificial intelligence. At that time, we said that if you want to design smart software, this software needs to have the following characteristics.
(1) Search capability.
(2) Language for knowledge representation.
(3) The ability to learn.
In early work, it was already obvious that blind search algorithms (i.e., no domain knowledge), such as breadth-first search and depth-first search algorithms, cannot effectively and successfully overcome the barrier to large-scale search space they face.
As mentioned in this book, a useful guideline is that if you want to design a system for performing a task, first check whether similar systems already exist in nature. If it is 1902 and you want to design a "flying machine", then your attention should be focused on birds. In 1903, the Wright brothers successfully built the aircraft. It is not surprising that the aircraft's fuselage is relatively thin and has two prominent large wings (see Figure 17.1).
Figure 17.1 This early model of Wright Brothers' aircraft presents a double-layer wing
blind search algorithm does not have the necessary functions to deal with large-scale search problems in the field of artificial intelligence. However, humans are expert-level "problem solving machines". Newell and Simon recognized this feature and studied humans who were asked to "think aloud" during the problem solving process. In 1957, this research eventually led to the invention of the General Problem Solver (GPS). The General Problem Solver has heuristics extracted from the human discipline and successfully solved the following problems: the kettle problem (see Chapter 1), the missionary and cannibal problems (see Chapter 2), and the Connisburg Bridge problem (see Chapter 6), etc. In the search algorithm in Chapter 3 and the game algorithm in Chapter 4 and Chapter 16, heuristics are effectively used, partially overcoming the problem of combinatorial explosion.
knowledge representation method also has a practical impact on the ability to solve problems. The Connisburg Bridge problem described in Chapter 6 is shown in Figure 6.6, and this picture is redrawn here, as shown in Figure 17.2.
The question is: "Can you pass these 7 bridges once and only once and return to the starting point?"
Figure 6.6 is a graphical model of the Connisburg Bridge. This part of the figure is re-drawn here, as shown in Figure 17.3.
Figure 17.2 Conniesburg Bridge
Figure 17.3 Conniesburg Bridge Graphic Model
1736, Leonhard Euler wrote his first paper on graph theory, giving the conclusion that if and only if the bridge shown in Figure 17.3 contains a ring and this ring contains all sides and vertices, the bridge shown in Figure 17.2 can traverse as described. Euler concluded that this graph contains such a ring (now called "Euler's ring") if and only if the degrees of each vertex are even.
Obviously, the representation of the problem has a huge impact on effectively discovering solutions. The above guidelines lead us to two learning paradigms. The human brain (and nervous system) is the most compelling example of natural learning systems. In Chapter 11, it emerges as a metaphor, in which we outline a learning approach—Artificial Neural Networks (ANN), which abstracts prominent features such as high connectivity, parallelism, and fault tolerance from human brain models. In many areas of problem solving, the ANN model can be considered successful, from economic forecasting to game and control systems.
The second paradigm is evolution, which may not be that obvious. Darwin (Darwin) describes how plants and animal species adapt to their environment and survive. Here, it is the species itself, not the individual who is learning. Chapter 12 outlines two evolutionary learning methods—Genetic Algorithm (GA) and Genetic Programming (GP). Both methods have been successful in the field of problem solving from scheduling to optimization.
17.2 Prometheus Return
In Greek mythology, Prometheus was a god who stole the fire from the heavens and brought the fire to the earth. Some accounts also gave him the heavy responsibility of creating humans from clay. In literature, the theme of creating life with inanimate materials is universal. Perhaps the most creepy description appears in the book Frankenstein or Mary Shelly's novel The Modern Prometheus. There is no doubt that readers are familiar with the story of scientists creating life and then terrifying at their creations. In 1931, Boris Karloff played the role of a monster in a film directed by James Whale. The first edition of the novel
Shelly was published in 1818, when the industrial revolution was in full swing. Humans have used steam power to carry out earth-shaking reforms in the manufacturing and textile industries. The invention of telegram has turned long-distance communication into instant communication. Many people believe that the aftermath of this revolution is not entirely beneficial. Our dependence on steam and coal-fired power, then oil, and recently nuclear energy has seriously polluted the planet, water, and air. Others believe that the Industrial Revolution promoted degenerate materialism. Literary critics point out very deeply that the morality of "Frankenstein" is that society must be wary of its attempts to control nature. As people's control over intelligent knowledge continues to increase throughout the 21st century, this may require the continuous emphasis on this warning to the artificial intelligence community.
One of the authors (S. L.) watched the movie in his childhood: but to this day, he still has the light on while sleeping.
Computer science is a field of science involving information and computing. The focus is on the algorithmic solution to the problem. The 20th century made this new student subject humble and cautious. As people discovered the basic limitations of problem solving, this discipline became more cautious. In other words, there may be some problems, and there are no algorithmic solutions to these problems. The famous problem is the so-called "halting problem". Given any process P, will P(w) pause when running any data w? For example, the four-color problem may be a well-known open-minded problem in graph theory. Its proposition is "Color the map, are the four colors enough to make the two adjacent areas different in colors?" In 1976, Appel and Haken gave a positive answer to this question. The computer program solved this problem for hundreds of hours.This will be of great benefit if the operating system running this program can predict that the program will eventually stop. Stop the question tells people that this prior knowledge is not always possible.
The book mentioned earlier Alan Turing (Allen Turing). In 1936, he was studying the problem of what kind of functions are computable. [3] For example, addition is a computable function, that is, a stepwise process can be given so that if the integers X and Y are taken as input, then after the finite calculation steps, their sum X + Y can be obtained. He provides a computing model now known as a Turing machine (see Figure 17.4). The Turing machine consists of the following three parts.
(1) Input/output tape, input problem on input/output tape writing; at the same time, the result is also written on the tape. Various Turing machine models exist; Figure 17.4 shows a model of two-way unbounded tape. The tape is divided into cells, and a symbol can be written in each cell. Each cell on the tape is preloaded with a blank symbol (B).
(2) A limited control containing algorithms (i.e., step-by-step process for solving problems).
(3) read/write header, which reads symbols on tape and writes symbols to this tape. It can be moved left or right.
Figure 17.4 Turing machine
Turing discusses the concept of a universal Turing machine (UTM) - this kind of Turing machine can run other Turing machines programs, that is, it can simulate the behavior of a "ordinary" Turing machine. Turing proves that for any Turing machine (T), it is impossible to determine whether the Turing machine (T) will stop at any input (w), i.e., T(w). This is the so-called Turing machine downtime problem. This more general version of the problem (i.e. downtime problem) cannot be proven to be undecided. People accepted this view without thinking, Turing-Chiao Qi's thesis gave. This paper mentions that Turing machines have similar computing power as digital computers. As a result, most computer scientists believe that problems that cannot be solved on Turing machines are also algorithmically unsolvable. Therefore, there are fundamental limitations in computing. As a subdiscipline of computer science, artificial intelligence also has these basic limitations. What one wants to know is whether the creation of human-level artificial intelligence also has these limitations.
17.3 Summary - Introduction to the achievements of artificial intelligence
In the subsequent chapters of this chapter, we return to the feasibility of creating human-level artificial intelligence. Now, we briefly introduce the achievements of artificial intelligence described in the first 16 chapters.
- search aspect. A* has been included in the
- video game design, which makes the game more realistic (see Chapter 3).
- Mapquest, Google and Yahoo maps use heuristic search. Many GPS and smartphone applications integrate this technology (see Chapter 3).
- uses the Hopfield network (see Chapter 11) and the evolutionary method (see Chapter 12) to find the approximate solutions to difficult, sometimes even NP-complete problems (such as TSP).
- game aspect.
- Minimax evaluation allows computers to play relatively simple games such as tic-tac-toe and nim (see Chapter 4).
- is assisted by heuristics and other machine learning tools, and through alpha-beta-trimmed Minimax evaluation allows computers to play tournament-level checkers (Samuels and Schaeffer) and chess (Deeper Blue beats world chess champion Garry Kasparov) (see Chapter 16).
- Championship level Othello program (Logistello, 1997), as well as "Mastered Players" in Backgammon (TD-Gammon, 1992), Bridge (Jack and WBridge 5, 2000s), and Poker (2007, see Chapter 16).
- fuzzy logic aspects.
- handheld camera automatically compensates for false hand movements.
- Automobile traction control device.
- Control devices for digital cameras, washing machines and other household appliances.
- Expert system aspects.
- has knowledge-intensive software or so-called expert system (ES) with built-in reasoning and interpretive devices, which can help consumers choose the right car model, browse online websites, shop, and more.
- ES can also be used for analysis, control, diagnosis (what disease does the patient have?), guidance and prediction (where should we dig oil?).
- ES is used in multiple fields such as drug, chemical analysis and computer configuration.
- As long as the ES system is used to help rather than replace humans, ES as one of the greatest achievements in the field of artificial intelligence will not be controversial (see Chapter 9).
- neural network aspect.
- Lexus cars have reversing cameras, sonar devices and neural networks. With these technologies, cars can be parked automatically in parallel.
- When the vehicle is too close to other vehicles or objects, Mercedes cars and other cars have automatic stop control.
- Google Cars are almost completely autonomous, but when it drives automatically, there must be someone in the car.
- Optical Character Reader (OCR) automatically routes large amounts of mail.
- automatic speech recognition system has been widely used. Software intelligence routinely helps us browse credit card and bank transactions.
- At the airport, the software provides automatic security alerts when people on the "no flight" list are detected.
- neural network assists in medical diagnosis and economic forecasting (see Chapter 11).
- evolution method.
- orbital scheduling of telecommunications satellites prevents communication from fading and disappearing.
- software for optimizing antenna and ultra-large scale integrated (VLSI) circuit design.
- data mining software makes data more valuable to the company (see Chapter 12).
- Natural Language Processing (NLP) aspect.
- conversational agent provides travel information to individuals and assists in booking hotel appointments, etc.
- GPS system usually sends voice commands to users, such as "at the next intersection, turn left." Some smartphones have applications that allow people to say requests: "Where is the recent coffee shop that can make cappuccinos?"
- Web request allows information retrieval across languages and language translation when needed.
- interactive agents provide verbal assistance to children who are learning to read (see Chapter 13).
- machine learning application with neural networks, natural language processing (see Chapter 13), voice understanding and planning (see Chapter 14), has made significant progress in robotics technology (see Chapter 15).
Overall, this is not a bad achievement for a computer science subdiscipline that began its second 50 years of development.
{App Window!}
GoogleDriving Cars
1998, Stanford graduate students Larry Page (Larry Page) and Sergey Brin founded Google. Google was originally a search engine called BackRub, which uses links to evaluate the importance of web pages. Google search engine is a joke about the word "googol", but it has achieved great success and has quickly become a powerful, well-known and mainstream search engine on the planet. Over the years, Google has also developed the same successful email system "Gmail" and the popular public video system "YouTube". Google also developed a driverless car. One of the engineers of
Google Driverless Cars (see Figure 17.5) is Dmitri Dolgov, who is headed by Dr. Sebastian Thrun. Thrun is a former director of the Artificial Intelligence Laboratory at Stanford University and is the co-inventor of Google Street View. Google driverless cars have been tested for several years and will continue to be presented as experimental in the coming years. While driverless cars seem to be a few years away from mass production, technicians believe they will be as popular as mobile phones and GPS systems in the near future. Google believes that this technology may not be profitable for many years, but in the possible sales of information and navigation services of other driverless car manufacturers, Google can foresee huge profits.
Google Driverless cars use artificial intelligence technologies such as laser dot marks sensing traces of anything nearby (such as marks and signs on the ground) to make decisions that human drivers should make, such as turning to avoid obstacles or stop when they see pedestrians.
According to the law, in order to prevent problems, there must be someone behind the steering wheel, and technicians are also required to monitor the navigation system to ensure the safety of the test and no accidents occur. For different drivers, you can choose different driving personalities such as "Driving with caution", "Driving with defensive driving" and "Driving with aggressive driving".
Robots usually respond faster than humans. Based on sensors and devices, the robot can fully sense it. It also won’t be distracted, nor will it have other factors that usually lead to accidents, such as fatigue, medications, and carelessness. The goal of engineers is to make these driverless cars more reliable than humans. Human error is the cause of many accidents. In addition, the software used by these driverless cars must be carefully tested and must be free of viruses and malware. Other concerns are fuel efficiency and space efficiency—that is, driverless cars will not have accidents in theory, so cars can be “crowded” on the road. Some Google driverless cars have a driving record of more than 1,600 kilometers, and there is no accident or human intervention. These vehicles have undergone a small number of human corrections and have a driving record of more than 100,000 kilometers. [1]
GoogleA test for driverless cars started outside campus near San Francisco. It uses various sensors within a range of about 182 meters and follows the route of the GPS compiled into the car. The car is traveling at a speed of about 105 kilometers per hour at a prescribed California speed. Just like humans, the car slows down when turning, and then accelerates a little bit later. The device located on top of the car provides a mapped version of the detailed environment and its surroundings, so it knows which roads need to be adopted, which roads to avoid, and which roads are dead ends. It can travel several miles on busy highways and can leave the highway without accidents. It can also drive through, park at red lights and stop signs, and be able to interact with pedestrians. If humans appear, it will wait for them to move. It has a voice system that announces its actions to the person or driver in the car. When an artificial intelligence system detects that there is a problem with the sensor, the driver will also be reminded. It also prevents accidents, using detection systems to point out what is happening. The driver can also regain control of the car by pressing the red button near the right hand, touching the brake or turning the steering wheel.
When the car is unmanned and the system automatically controls it, this is called Cruise mode (cruise mode). At this time, people in the car can release the steering wheel. In fact, it has become a public transportation with no fees, no crowds, no gaze and no other factors (those can distract the average car driver).
However, there are still some legal issues, such as if an accident occurs, who will be responsible for it. All states that allow self-driving vehicle testing have no relevant laws on the situation where accidents occur while driving a driverless car. Google found that as long as there is someone in the vehicle of the driverless car and this person can control any possible error, it is legal to drive a driverless car.
Google Driverless cars will reduce the demand for private cars, thereby reducing traffic flow, allowing people to have more available land without the need to pave roads more widely.
Recently, Google has been building experimental electric vehicles with normal control standards that do not require driver control except starting and stopping the vehicle. People can command the car to automatically drive through a smartphone app to reach the location of people who need it and take people to their destination.The car also invented a feature, the so-called Traffic Jam Assist function, which allows driverless cars to follow another car while driving.
Google’s plan for driverless cars is to have at least 100 new prototype cars powered by electricity. Google's team will limit them to travel in urban and suburbs at about 40 km/h. The test will be conducted by Google personnel, which will help with testing in small and closed areas. Naturally, it takes some time to convince regulators that it is safe for people to use driverless cars.
Figure 17.5 Google Driverless Car
Reference
Thrun S. What we’re driving at. Google, 2010.
Markoff J. Google’s next phase in driverless cars: No steering wheel or brake pedals. New York Times, 2009.
Markoff J. Google Cars drive themselves, in traffic. New York Times, 2014.
17.4 IBM's Watson-Danger Edge Challenge
People and machines battles provide a system that inspires people's enthusiasm and publicity for some technological achievements. IBM is the initiator of these three events. The first incident occurred in 1997, when Deeper Blue, a parallel computer with a special purpose and search device, defeated the world champion of chess in six matches (see Chapter 16).
A TFLOP (teraflop) represents one trillion (1012) floating point operations per second.
Blue Gene is a project that focuses on the production of some high-speed supercomputers to study biomolecular phenomena. The machine of this project has achieved speeds of hundreds of TFLOPs. In 2014, the Blue Gene/L system exceeded 36 trillion times per second.
A petaflop corresponds to one thousand trillion (1015) floating point operations per second.
In the past few years, the Watson-Danger Edge Challenge of IBM has been underway. The goal is to design a computer that can answer questions raised in natural language, which is full of ambiguity. In the field of natural language processing, question-and-answer systems are not new (see Chapter 13). However, IBM hopes that Watson can perform at a speed comparable to excellent human players (2-3 seconds).
Information about IBM's Watson-Hazardous Edge Challenge can be found on the web. Enter "www.IBM.com" and "Watson-Jeopardy Challenge" in turn.
Top human contestants have the information on numerous different topics that encompass everything from world geography to Broadway drama, literature to pop culture. Some existing problems with
are as follows.
(1) "In 2000, the 100th episode of Got Milk advertisement shows a pop singer who looks like 3 and 18 years old, who is she?" The correct answer is: "Britney Spears" Blue J (Watson's early name) answered: "Holy Crap".
(2) "In a nine-ball game, whenever you hit a ball into the bag, you have to start over." Blue J's answer was correct: "Cueball".
(3) "Which country shares the longest boundary with Chile ?" Blue J's answer is incorrect: "Bolivia" The correct answer is the second option " Argentina ".
In 2007, David Ferrucci, a senior employee of IBM, was selected as the head of the Watson development team. He has extensive experience in language processing systems. In Stephen Baker's bestseller [4], Ferrucci admits two conflicting fears: the first is that Watson (and IBM) has suffered a crushing defeat on the national stage after years and millions of dollars in research; the second is that at the last minute, another company will bypass IBM and design a winning system. As it turns out, these fears have accompanied him for four years. Ferrucci understands that if Watson is going to succeed, it has to load facts—not just facts, but the right facts.So they studied and classified thousands of past Jeopardy issues and decided to get Watson to load "tons" of Wikipedia articles. Next, Watson downloaded the Gutenberg Library and "study" the works of famous artists. Watson also gathers insights from human competitors. Early in the Watson project, it was discovered that deep knowledge was not necessary—traditional knowledge with many different topics was sufficient. Instead of preparing for the competition by studying several thick books, Ken Jennings practiced with flash cards, hoping to have some superficial knowledge on a wide range of topics.
Next, the developer fed Watson encyclopedia, dictionary, news articles and web pages in a cramming manner. As Baker describes: "(Warson) was painful and slow." In the following years, Watson began playing against former Dangerous Fringe contenders. Slowly, its performance began to improve.
Watson consists of more than 2000 processors, each working in parallel, following different inference threads. It shows several answers for each question and lists the confidence of each answer. Whenever Watson was confident in one of the answers, it quickly pressed the buzzer.
Gradually, in the face of human competition, Watson began to perform well. It occasionally loses its words and makes profane words. Of course, the corporate image of IBM is important; they installed a filter so that Watson would not issue the most common profane language. The
man-machine competition was held in early March 2011. Despite some awkward mistakes, Watson eventually won.其中最有名的失误是最后一道危险边缘问题：
“它最大的机场是以第二次世界大战来命名。”在“美国城市”的类别中，沃森回答说：“多伦多”
为了给沃森辩护，Ferrucci解释说，伊利诺伊州有一个多伦多，多伦多也拥有一支美国职业棒球队。 However, the fact is that Watson made a mistake. Of course, an interesting question is: "What kind of future does a Watson-like machine have?" Dangerous Edge Champion Computers certainly have no market. However, IBM expects that Watson and his successors will receive expert training in the fields of medicine, law, etc., where new knowledge is being discovered at an astonishing speed. This would be of great benefit if “Medical Watson” reads the latest journal and can advise doctors on the best treatments for patients. Alternatively, “Law Watson” can identify precedents and find the defense point of the law.
To help promote the Watson-Danger Edge Challenge, IBM sent representatives to City College of New York and CUNY Graduate School (CUNY) in February 2011. Wlodek Zadrozny, one of the team members of the IBM, gave a lecture at City College in New York. The IBM team members participating in this event are shown in Figure 17.6. Figure 17.7 shows Wlodek Zadrozny discussing Watson with attendees at City College of New York. Finally, Jerry Moy hosted two CUNY demonstrations, as shown in Figure 17.8.
图17.6 在纽约城市学院的IBM团队成员（左至右）：Bruno Bonetti、Jerry Moy、Faton Avdiu、Arif Sheikh、Andrew Rosenberg、Wlodek Zadrozny、Raul Fernandez、Vincent DiPalermo、Andy Aaron和Rolando Franco
图17.7 Wlodek Zadrozny与纽约城市学院的与会者一起讨论沃森
图17.8 Jerry Moy hosted two CUNY demonstrations.
. The book often mentions that the correct role of artificial intelligence technology is to assist humans, not replace humans. Watson will provide valuable help to human experts in different fields.
{人物轶事}
雷·库兹维尔（Ray Kurzweil）
Ray Kurzweil（见图17.9）是世界著名的科学家、发明家、企业家和未来学家。 Forbes magazine called him "the legal heir to Thomas Edison " and listed him as one of the eight top entrepreneurs in the world. It has always been said that Kurzweil “has become an industry in its own right."Some of his famous inventions include the first CCD flatbed scanner, the first all-round font optical character recognition, the first blind print voice reader, the first text speech synthesizer, the first music synthesizer (can reproduce the grand piano and other orchestral instruments) and the large vocabulary speech recognition system sold on the market.
Figure 17.9 Ray Kurzweil
Kurzweil received a $500,000 MIT-Lemelson Award, a grand prize for innovation. In 1999, he received the National Medal of Technology, the highest national honor in the United States in terms of technology. In 2002, he officially entered the National Inventors Hall of Fame established by the U.S. Patent Office.
In addition, he received 20 honorary doctorates, with three US presidents awarded it. He created 7 books, 5 of which are bestsellers. The Age of Spriritual Machines has been translated into 9 languages and was ranked first in the bestseller of Amazon science. His book The Singularity Is Near is a New York Times bestseller and was the number one book on Amazon in science and philosophy.
In 2012, Kurzweil was appointed director of engineering at Google, leading the team to develop machine intelligence and natural language processing. Kurzweil's books also include:
- The Age of Intelligent Machines (1990).
- The 10% Solution for a Healthy Life (1993).
- The Age of Spiritual Machines (1999).
- Fantastic Voyage (with Dr. Terry Grossman) (2004).
- The Singularity (2005).
- Transcend: Nine Steps to Living Well (co-authored with Dr. Terry Grossman) (2009).
- "How to Create a Mind" (2012).
Most of the information about Ray Kurzweil here comes from the Kurzweil AI website.
Singularity
In 2005, Ray Kurzweil published the book "The Singularity is Near: When Humans Transcend Biology", which is probably the most controversial book he has published. The central theme of this magnificent work is what he calls “Law of Accelerating Returns.” He believes that computers, genetics, nanotechnology and artificial intelligence are growing exponentially. According to him, by 2045, artificial intelligence will surpass the human intelligence on this planet. Figure 17.10 is the singularity depicted by the Kurzweil AI homepage.
Figure 17.10 The singularity depicted by the home page of KurzweilAI.net
Kuzweil believes that evolution must go through the following 6 stages:
(1) Physics and Chemistry.
(2) Biology and DNA.
(3) Brain.
(4) technology.
(5) The fusion of human technology and human intelligence.
(6) The universe wakes up.
He claimed that the first 4 stages have occurred, and humanity is now in the fifth stage. By 2045, technology will make rapid progress and people will be able to make their bodies healthier through nanotechnology and artificial intelligence. The Moore's Law described by
Kurzweil.net is shown in Figure 17.11.
Figure 17.11 Moore's Law described in Kurzweil.net
17.5 Artificial Intelligence in the 21st Century
Return to the unanswered question raised in the previous discussion: Will the creation of human-level artificial intelligence exceed the basic boundaries of artificial intelligence? Let’s first think about the origin of human intelligence, and then think about the origin of life itself.
The famous British scientist Richard Dawkins (Richard Dawkins) [5] solved the latter problem, and he found insights in Darwin's theory of evolution. Of course, 4 billion years ago, there were no animals or plants on Earth—just the "primitive soup" of basic atoms.Dawkins believes that Darwin's theory can be generalized to "stabilizer survival", in other words, stable atoms (and molecules) are more likely to survive on this ancient earth. He further speculated that in early history, the planet was rich in water, carbon dioxide, methane and ammonia, and therefore could form amino acids (complex molecules that are components of proteins). Proteins are known precursors of life. Dawkins envisions that on the long road to life on this planet, the next step is the unexpected creation of the so-called "replication factor". This replication factor has a significant property - being able to faithfully replicate itself. He believes that in this primitive environment, it is stable to be able to quickly and accurately copy one's own replication factor.
replication (or reproduction) process itself requires a stable supply of basic "raw materials". There is no doubt that different replication factors continue to compete to obtain a full supply of water, carbon dioxide, methane and ammonia. This evolution process lasted for 4 billion years. Dawkins believes that after this long evolutionary round, among the animals and plants that live on this planet today, we can find the successor - this is the gene.
Regarding possible origins of life on this planet, Dawkins continues its extraordinary discourse by explaining how these genes work to ensure survival. For the past 600 million years, they have acted very much like the fictional elves quoted in Chapter 12. They have been shaping human eyes, ears, lungs, etc., and the boat of life (i.e. the body) is built from these organs. In this discussion, the animal's body and plants seem to be just a protective partition that protects the survival of all important genes. Recently, with an in-depth (SL) reading of Dawkins' work, my mind returned to a scene from the Star Wars movie series. In this scene, enemy troops place soldiers in robotic combat machines equipped with giant legs, which forms the soldier's protective shell. Even if we accept Dawkins' theory, there is a question - "Where is the origin of human consciousness?" Dawkins might think that animals with consciousness (agained by natural selection) will have advantages and thus achieve relative stability, thus ensuring survival.
Gerald Edelman is a biologist who has won the Nobel Prize. He proposed a theory of consciousness biology [6], which is also based on Darwinism. He believes that consciousness and mind are purely physiological phenomena. Neuron groups self-organize into many complex and highly adaptable modules. Edelman believes that the brain is functionally plastic, that is, because the human genome does not have enough coding capabilities to completely specify brain structures, a large number of brain tissues are self-oriented.
In physics, unified field theory should be a theory about everything. This theory tries to unify various forces occurring in nature, such as gravity, electromagnetic force, strong force and weak force.
Marvin Minsky solved a broader problem in Society of Mind [7]. He asked, “How does the brain organize?” “How does cognition happen?” As Dawkins tells us, the human brain has evolved over hundreds of millions of years. The unified theory cannot simply and bluntly explain the functions of complex organs in human skulls. Building a kind of wisdom is like forming an orchestra without a conductor. Among them, musical instruments are agents (see Chapter 6), and they are not playing music, but explaining the world. Some agents help to understand language, others can explain visual scenes, and others provide common sense for humans (see discussion of the Cyc project in Chapter 9). Unless effective communication can be conducted between agents, all this makes no sense. Minsky hypothesized that at any point in time, an individual's psychological state can be interpreted as a function in which a subset of agents is active. Perhaps artificial intelligence is still a field that is too young and is not ready to propose an intelligent "unified field theory" like Minsky. But when AI matures, Minsky's Society of Mind may play a prominent role in it.
2015, at the biological and chemical level, people fully understand the functions of individual neurons. In human knowledge, the still shortcomings are how a group of neurons process sensory data, encode experience, understand language, and in a more general sense how to promote cognition and initiate consciousness. Current research uses X-rays and other scanning techniques to gain understanding of the brain at the functional module level. Kurzweil predicts that by the mid-21st century, we will have a complete, architectural understanding of the human brain.
In addition, he speculates that miniaturization of computer components will advance to a new stage, by which time it is feasible to use hardware to fully implement the brain—this implementation may require connections between billions of artificial neurons and trillions of even billions of neurons. Perhaps at that time, we will have enough power to achieve artificial intelligence at the human level. It is wiser for us to remember that Prometheus created the "reward" of fully conscious human beings, that is, he is tied up so that the lion can enjoy his liver, and then his liver regenerates, allowing the lion to enjoy his liver again. Science fiction literature outlines countless scenarios in which humans create artificial intelligence at the human level. We hope that if AI can always follow this lofty goal, this reward will be more satisfying than the “reward” given to Prometheus.
17.6 Summary of this chapter
In this chapter, we review many achievements in the field of artificial intelligence over the past 50 years. We put artificial intelligence in one framework—as a subdiscipline of computer science. Just as it is undecided to consider the well-known downtime problem in computer science, we also think about the proposition of "whether it is possible to create a human-level artificial intelligence." We discuss the Watson system of IBM and describe its capabilities as an assistant to legal and medical professionals.
Finally, by thinking about the origins of life, intelligence and consciousness, we summarize and introduce Kurzweil's optimistic view of the possibility of successfully creating artificial intelligence at the human level in the near future.
This article is excerpted from "Artificial Intelligence" (2nd Edition)
Artificial Intelligence (2nd Edition)
[US] Stephen Lucci, Danny Kopec, written by
- Artificial Intelligence Encyclopedia
- Easy to use Artificial Intelligence Self-study Guide
- covers machine learning, deep learning, natural language processing, neural networks, computer games and other knowledge, detailed graphics and texts, detailed explanations, rich teaching resources and learning materials,
- American classic textbooks, in the United States, was evaluated as Russell Norvig's "Artificial Intelligence: A Modern Method" will be better for undergraduates.
This book is an artificial intelligence textbook carefully written by the author based on his years of teaching experience. It can be called the "encyclopedia of artificial intelligence". The book covers topics such as a brief history of artificial intelligence, search methods, informed search, search in game, logic in artificial intelligence, knowledge representation, production systems, expert systems, machine learning and neural networks, genetic algorithms, natural language processing, automatic planning, robotics, advanced computer games, the history and future of artificial intelligence.
This book provides rich teaching supporting resources, suitable as a textbook for artificial intelligence-related majors in higher education institutions, and is also suitable for readers who are interested in artificial intelligence-related fields.
