On April 27th, the famous physicist stephen hawking gave a video speech at the Global Mobile Internet Conference held in Beijing. In his speech, Hawking reiterated that the rise of artificial intelligence is either the best thing or the worst thing for human beings. He believes that human beings need to be alert to the threat of artificial intelligence development. Because once the artificial intelligence breaks away from the bondage and redesigns itself in an accelerating state, human beings will not be able to compete with it because of the limitation of long biological evolution, thus being replaced.
The following is Hawking’s speech:
In my life, I have witnessed profound changes in society. One of the most profound changes, which has an increasing influence on human beings, is the rise of artificial intelligence. To put it simply, I think the rise of powerful artificial intelligence is either the best thing or the worst thing in human history. I have to say, whether it is good or bad is still uncertain. But we should do our best to ensure that its future development is beneficial to us and our environment. We have no choice. I think the development of artificial intelligence itself is a trend with problems, and these problems must be solved now and in the future.
The research and development of artificial intelligence is advancing rapidly. Maybe all of us should pause for a moment and shift our research duplication from improving the ability of artificial intelligence to maximizing the social benefits of artificial intelligence. Based on this consideration, american association for artificial intelligence (AAAI) set up a forum on the long-term future of artificial intelligence in 2008-2009, and they have recently invested a lot of attention in the purpose-oriented neutral technology. But our artificial intelligence system needs to work according to our will. Interdisciplinary research is a possible way forward: it extends from economics, law and philosophy to computer security, formal methods and, of course, various branches of artificial intelligence itself.
Everything produced by civilization is the product of human intelligence. I believe that there is no essential difference between what the biological brain can achieve and what the computer can achieve. Therefore, it follows the principle that computers can imitate human intelligence in theory and then surpass it. But we are not sure, so we can’t know whether we will be helped by artificial intelligence indefinitely, despised and marginalized, or probably destroyed by it. Indeed, we are worried that smart machines will be able to replace the jobs that human beings are doing and quickly eliminate millions of jobs.
While artificial intelligence has been developing from primitive form and proved to be very useful, I am also worried about the result of creating something that can equal or surpass human beings: once artificial intelligence is out of bondage, it will redesign itself in an accelerating state. Limited by the long biological evolution, human beings cannot compete with it and will be replaced. This will bring great damage to our economy. In the future, artificial intelligence can develop from my will, a will that conflicts with us. Although I have always been optimistic about human beings, others believe that human beings can control the development of technology for a long time, so that we can see the potential of artificial intelligence to solve most problems in the world. But I’m not sure.
While artificial intelligence has been developing from primitive form and proved to be very useful, I am also worried about the result of creating something that can equal or surpass human beings: once artificial intelligence is out of bondage, it will redesign itself in an accelerating state. Limited by the long biological evolution, human beings cannot compete with it and will be replaced. This will bring great damage to our economy. In the future, artificial intelligence can develop from my will, a will that conflicts with us. Although I have always been optimistic about human beings, others believe that human beings can control the development of technology for a long time, so that we can see the potential of artificial intelligence to solve most problems in the world. But I’m not sure.
In January 2015, I signed an open letter on artificial intelligence with elon musk, a technology entrepreneur, and many other artificial intelligence experts, with the aim of advocating serious research on the impact of artificial intelligence on society. Before this, elon musk warned people that superhuman artificial intelligence may bring immeasurable benefits, but if it is not deployed properly, it may bring the opposite effect to human beings. I am with him on the Scientific Advisory Committee of the Institute for the Future of Life, an organization to alleviate the existing risks faced by mankind, and the aforementioned open letter was also drafted by this organization. This open letter calls for direct research that can prevent potential problems, at the same time, it also gains the potential benefits brought by artificial intelligence, and at the same time, it is committed to making artificial intelligence researchers pay more attention to artificial intelligence security. In addition, for policy makers and the general public, this open letter is informative and not alarmist. Everyone knows that artificial intelligence researchers are seriously thinking about these concerns and ethical issues, and we think this is very important. For example, artificial intelligence has the potential to eradicate diseases and poverty, but researchers must be able to create controllable artificial intelligence. The open letter with only four paragraphs entitled "Priority should be given to the study of powerful and beneficial artificial intelligence" made detailed arrangements for the research priorities in its attached 12-page document.
In the past 20 years, artificial intelligence has been focusing on the problems arising from the construction of intelligent agents, that is, various systems that can perceive and act in specific environments. In this case, intelligence is a rational concept related to statistics and economics. Generally speaking, this is the ability to make good decisions, plans and inferences. Based on these works, a large number of integration and cross-breeding have been applied in artificial intelligence, machine learning, statistics, cybernetics, neuroscience, and other fields. The establishment of sharing theoretical framework, combined with the ability of data supply and processing, has achieved remarkable success in various subdivided fields. Such as speech recognition, image classification, automatic driving, machine translation, gait movement and question answering system.
With the development of these fields, a virtuous circle is formed from laboratory research to economically valuable technologies. Even a small performance improvement will bring huge economic benefits, and then encourage longer-term and greater investment and research. At present, it is widely recognized that the research of artificial intelligence is developing steadily, and its influence on society is likely to expand, and its potential benefits are enormous. Since everything produced by civilization is the product of human intelligence; We can’t predict what we may achieve when this intelligence is amplified by artificial intelligence tools. However, as I said, it is not impossible to eradicate disease and poverty. Because of the great potential of artificial intelligence, it is very important to study how to benefit and avoid risks.
Now, the research on artificial intelligence is developing rapidly. This study can be discussed in the short and long term. Some short-term concerns are about driverless driving, from civilian drones to autonomous vehicles. For example, in an emergency, a driverless car has to choose between a big accident with small risk and a small accident with high probability. Another concern is lethal intelligent autonomous weapons. Should they be banned? If so, how to define "autonomy" accurately. If not, how to be accountable for any improper use and fault. There are other concerns. Artificial intelligence can gradually interpret the privacy and concerns caused by a large number of monitoring data, and how to manage the economic impact caused by artificial intelligence replacing jobs.
The long-term concern is mainly the potential risk that the artificial intelligence system is out of control. With the rise of super intelligence that does not follow human wishes, that powerful system threatens human beings. Is it possible to have such a misplaced result? If so, how did these situations arise? What kind of research should we invest in order to better understand and solve the possibility of the rise of dangerous super intelligence, or the emergence of intelligent outbreaks?
At present, the tools to control artificial intelligence technology, such as reinforcement learning and simple and practical functions, are not enough to solve this problem. Therefore, we need further research to find and confirm a reliable solution to control this problem.
Recent milestones, such as the self-driving car mentioned earlier and the artificial intelligence winning the Go game, are all signs of future trends. Huge investment has been poured into this technology. What we have achieved now is bound to be dwarfed by what we may achieve in the next few decades. And we are far from predicting what we can achieve when our minds are magnified by artificial intelligence. Perhaps with the help of this new technological revolution, we can solve the damage caused by some industrialization to nature. All aspects of our lives are about to be changed. In short, the success of artificial intelligence may be the biggest event in the history of human civilization.
But artificial intelligence may also be the end of human civilization, unless we learn how to avoid danger. I once said that the all-round development of artificial intelligence may lead to the extinction of mankind, such as maximizing the use of intelligent autonomous weapons. Earlier this year, I and some scientists from all over the world supported its ban on nuclear weapons at the United Nations meeting. We are anxiously awaiting the result of the negotiation. At present, nine nuclear powers can control about 14,000 nuclear weapons, and any one of them can raze cities. Radioactive waste will pollute farmland in a large area, and the most terrible harm is to induce nuclear winter. Fire and smoke will lead to the global Little Ice Age. This result has caused the global food system to collapse, with doomsday turmoil, which is likely to lead to the death of most people. As scientists, we bear a special responsibility for nuclear weapons, because it is scientists who invented them and found that their influence is more terrible than originally expected.
At this stage, my discussion of disasters may have frightened everyone present. I’m sorry. But as today’s participants, it is important for you to recognize your position in the future research and development that will affect the current technology. I believe that we are United to call for the support of international treaties or sign open letters to governments of various countries, and scientific and technological leaders and scientists are doing their utmost to avoid the uncontrollable rise of artificial intelligence.
Last October, I set up a new organization in Cambridge, England, trying to solve some inconclusive problems in the rapid development of artificial intelligence research. "Livelihoom Intelligent Future Center" is an interdisciplinary research institute dedicated to studying the intelligent future, which is crucial to the future of our civilization and species. We spend a lot of time studying history and going deep into it — — Mostly about stupid history. So it is a gratifying change that people turn to study the future of intelligence. Although we are aware of the potential dangers, I am still optimistic, and I believe that the potential benefits of creating intelligence are enormous. Perhaps with the help of this tool of new technological revolution, we will be able to reduce the harm caused by industrialization to nature.
Every aspect of our lives will be changed Hugh Prince, my colleague in the graduate school, admitted that the Livelihoom Center was established in part because the university established the Center for Existential Risk. The latter examines the potential problems of human beings more widely, while the focus of the Livelihood Center is relatively narrow.
The latest development of artificial intelligence, including the European Parliament’s call for drafting a series of regulations to manage the innovation of robots and artificial intelligence. Surprisingly, there is a form of electronic personality involved to ensure the rights and responsibilities of the most capable and advanced artificial intelligence. A spokesman for the European Parliament commented that as more and more fields in daily life are increasingly influenced by robots, we need to ensure that robots serve mankind now and in the future. The report submitted to members of the European Parliament clearly believes that the world is at the forefront of the new industrial robot revolution. It may be possible to analyze whether to provide robots with the rights as electronic people, which is equivalent to legal persons. The report emphasizes that at all times, researchers and designers should ensure that every robot design contains a stop switch. In Kubrick’s movie 2001: A Space Odyssey, Hal, a malfunctioning supercomputer, didn’t let scientists into the space capsule, but it was science fiction. What we have to face is the truth. Lorna Blazere, a partner of Osborne Clark International Law Firm, said in the report that we don’t recognize whales and gorillas have personalities, so there is no need to rush to accept a robot personality. But concerns have always existed. The report admits that in a few decades, artificial intelligence may surpass the scope of human intelligence, and artificial intelligence may surpass the scope of human intelligence, thus challenging human-computer relations. The report finally calls for the establishment of European robotics and artificial intelligence institutions.To provide technical, ethical and regulatory expertise. If members of the European Parliament vote for legislation, the report will be submitted to the European Commission. It will decide which legislative steps to take within three months.
We should also play a role in ensuring that the next generation not only has the opportunity but also has the determination to fully participate in scientific research at an early stage, so that they can continue to exert their potential and help mankind create a better world. This is what I meant when I first talked about the importance of learning and education. We need to jump out of the theoretical discussion of "how things should be" and take action to ensure that they have the opportunity to participate. We are standing at the entrance to a beautiful new world. This is an exciting and uncertain world, and you are the pioneers. I bless you.
