Why Elon Musk and we are "afraid" of artificial intelligence?

Before I start writing on a topic, I usually look up the information and make a list of outlines. And when it comes to writing about AI, a concept that has recently been fired up by a lot of startups getting funding, I think of something interesting.

Artificial intelligence (Artificial Intelligence,ai) is an ancient, and has not been the concept of too many people's attention in the past 50 years. Read Wikipedia, you can even find "ai winter" this term. In fact, the superficial knowledge of artificial intelligence does not create any hindrance to people's hopes of discussion under the framework of this topic--I am the best example: in "ai In the winter, the general public and the general public, who have been in the second half of the last century, have not made much of an understanding of AI, but in the past people were more willing to mention "machine learning" (Machine Learning), and AI became more popular in recent years.

If you are not an algorithmic engineer, then you do not need to understand how artificial intelligence is generated, just know it is there, it is so easy to use ... Wait a minute.

My logic was interrupted here and then knocked down.

During the "ai winter, artificial intelligence research stalled with the failure of several technological innovations, and the machine learning of the changed names was more likely to be favored by investors. Machine learning and artificial intelligence research methods are interlinked, but the goals are very different: machine learning wants to imitate humans, and artificial intelligence hopes to replace humans. Like self-driving and unmanned, this is a highly comparable degree of automation and intelligence, but there are still fundamentally different vehicle control technologies. After my logic has been crushed by myself, the logical reconstruction will revolve around autopilot and unmanned driving:

One months ago, Tesla, the world's hottest electric car company, "d" the theme of the press conference, attracting the attention of technology fans and car enthusiasts worldwide. Half a month later, people finally know what "d" stands for, but the hope of unmanned driving this rich future fantasy features, or did not come to Tesla's new car, so that people can not help but some disappointment. Before that, Google, another Silicon Valley technology company, had just released a five-year-old unmanned car, with no steering wheel, no throttle, no brakes, and other google[x.

Tesla is a car company and Google is ... Search the company. Why should we rely on a search company, not a car company, to guide the future of unmanned driving? What is blocking the world's most high-tech talent-intensive Silicon Valley, with the world's best automotive-related technology, with almost endless capital to invest in the development of Tesla, to do driverless cars? Elon Musk, founder and CEO of Tesla, who has recently made public comments on several occasions, may be able to answer our questions:

1. Musk previously accepted the Chinese media "blog World" interview, alone with an interview team three test drive their own development of electric vehicles. Alarmed by the superior performance of Tesla's new car, the interviewer wants to know how Musk feels about driving his own car. Musk's response is as high as usual:

We try to make the operation of this car more intuitive to the driver, as if the car is a part of itself, control the car as if you control your body-no need to think, the car is there, ready to sense your ideas, do what you want to do.

2. As early as August this year, Musk had expressed doubts about artificial intelligence on Twitter:

Nick Bostrom's book "Super Smart" (superintellegence) is worth seeing. We need to be very cautious about AI, otherwise it will be as dangerous as nuclear weapons in the future.

Not every tweet of a tech company leader is worth hammering at the time, Musk. But not too long, Musk in a speech once again that he is not talking about playing.

3. In an open interview with MIT, October 26, Musk said that artificial intelligence, like the "summon the demon" myth that ancient warlocks would study, would have unpredictable consequences for those who studied it and used it:

The study of artificial intelligence is like summoning demons. Like all the mythical wizards of the Holy water, every sorcerer claims he can control the demon, but none of them finally succeeds.

It is surprising that such an "anti-innovation" statement could be made by a senior scientific and technological person with a good technical background and rich project experience, not to mention that he had just been with a start-up company that studied image recognition technology in the field of artificial intelligence this March.

But from a different perspective, these comments may be better understood: Musk's understanding of Tesla electric cars is that "controlling the car is just like controlling your body-no need to think, the car is there, ready to sense your thoughts, do what you want to do", which accurately embodies Musk's belief that human nature should have control over the car. , not vice versa. And precisely because Musk is a technical man, he has more data than the average man can think of in the present and foreseeable future, the extent to which artificial intelligence has replaced/enslaved humans. Originally, he was touted as the real version of Iron Man, the most concerned about the excessive development of artificial intelligence technology.

What can AI do now? A company called Darktrace is using Bayesian theory to develop a security system that attempts to learn what is "normal" and "abnormal", thus providing an IT infrastructure for deploying the system to provide an incident or even a warning service before it occurs; another named Celaton Company has been able to provide users with virtual customer service software, through continuous learning, to help users in customer service work of this labor-intensive work, reduce labor costs, improve customer service communication effectiveness.

Dark blue, developed by human beings, has defeated human beings in chess.

One side is Tesla as an innovative electric vehicle company, in the automatic driving system above the "no pursuit", the other side is a good artificial intelligence start-ups hope that through their own technology, using machines to replace humans, to bring about great changes in production efficiency. Why did Tesla's new car not open more power to the machine system, but only gave it an upgraded version of the past petrol-car cruise: Intelligent acceleration, deceleration, and so on? Why not try to make the new car a complete replacement for human control? The lack of machine scanning and mapping of roads that cannot support unmanned driving is one reason, in terms of realistic constraints. and worry that the machine system will one day go to "The Singularity of Information", the integration of the human psychological model to abandon the three laws set by mankind, the choice of betrayal of human masters, perhaps more important reason.

Autopilot wants to optimize driving experience the goal of unmanned driving is to replace the driver; The ultimate goal of machine learning is to integrate data and computing power into a cloud-like calculator-still in use, and AI researchers may not be able to confirm their own research systems, At the peak of high wisdom, will choose to replace the emotional, life cycle, inefficient human beings. When the machine is ready to decide what is "normal" and "abnormal", can let the communication object realize oneself is the machine, facing the machine is very likely to abandon "the loyalty" this greatly affects the machine system own efficiency behavior, has become more and more lazy mankind still has how big odds?

Countless outstanding literature, film and television works have been interpreted above the occurrence and prosperity. By then, the machine is not a human calculator, and human beings will become the battery of the machine. Some of my friends have had this problem: human research on artificial intelligence, really to the point of worry about it? Perhaps Elon Musk wants to convey to us that this is not the case. Indeed, a modest artificial intelligence study can bring progress to human life, but I'm afraid no one has yet been able to figure out how much this "degree" is, and to say it, to convince ordinary people like me to have visions, awe and fear of AI. This is the most frightening point of artificial intelligence, like the autopilot and drone relationships in the logical reconstruction of this article: how many people in the world would be willing to believe that driverless cars would not have killed themselves without steering wheel and braking equipment?

Before the proper "degree" of AI research is found, no matter how emotional, inefficient or backward, human mastery of control is still the only and best choice.

(Responsible editor: Mengyishan)