Does artificial intelligence protect privacy?

From 65438 to 0997, Deep Blue defeated the top chess players many times, which attracted great attention all over the world. People are both excited and worried about the trend that artificial intelligence continues to surpass human intelligence. Will artificial intelligence infringe on human interests while making great contributions to human society? This has become the main concern of many scientists, politicians and the public.

So, Isaac, the master of science popularization and science fiction? Asimov's three laws of robots have been widely mentioned and discussed. The first rule: robots must not harm human beings, or stand by when human beings are hurt; The second law: unless the first law is violated, the robot must obey the orders of human beings; Law 3: Robots must protect themselves without violating the first and second laws.

After the robot was extended to artificial intelligence, this "three laws" was generally recognized and even became the research and development principle of artificial intelligence.

Today, 20 years later, Alpha Dog has defeated many international Go masters, but people are generally no longer worried, but are happy to see the complete victory of artificial intelligence.

Why do people's attitudes change? It is not that the potential harm of artificial intelligence does not exist, but that people have long been accustomed to it.

In the past two decades, with the vigorous development of mobile Internet, everyone has become accustomed to the status quo, that is, each of us is transparent in the information society.

The eye surveillance camera on the street recorded all your activities. In the monitoring hall of public security organs, through the linkage of multiple cameras, your walking track can be seen at a glance.

Access to airports, docks, stations, and even check-in information are completely recorded, and privacy in the traditional sense is no longer established. Buy a house, buy a car, and enroll your children in school. Your registered personal information has already been leaked, and now it is being used by fraudsters. People have long adapted to the information society where privacy is gradually deprived, but what can they do if they don't adapt?

But the essence of artificial intelligence is to serve human society, and the core value of human society is "people-oriented". Privacy is the eternal basic right of human beings, and the development of artificial intelligence cannot be at the expense of depriving human beings of their basic rights. On the contrary, artificial intelligence should better protect human basic rights, including privacy, which is the correct direction of artificial intelligence development.

In the era of mobile internet, the information of ordinary users can be divided into two categories, one is identity authentication information, such as our user name and login password, as well as biometric passwords such as fingerprints and irises, and the other is our content information, such as ordering food, taking a taxi, shopping and renting a house.

Identity authentication information must be absolutely controlled by the user himself and cannot be transferred to any merchant or even the government. This should be an iron law of artificial intelligence.

Why do you say that? Because in the era of mobile internet, once your identity authentication information is leaked and used by the bad guys, it may suffer huge economic losses and even endanger your life. You must never trust any organization or individual in such an important matter.

We can't trust not only their honesty, but also their ability. In recent years, many well-known websites have leaked user authentication information. These enterprises are honest, but their protection ability is not enough, and they have been copied by hackers.

What about the content information of ordering food, taking a taxi, shopping and renting a house? Unlike authentication information, this part of information cannot be completely in your own hands. In order to enjoy the service of artificial intelligence, you sometimes have to give up.

Artificial intelligence has four links: intelligent perception, intelligent reasoning, intelligent learning and intelligent action, and these links are all driven by data. If you don't provide it with data, how can it understand your needs? How to provide you with accurate services through learning and reasoning?

Although the content information must be transferred, we must also adhere to the principle of "the statistical results of data belong to the merchants and the ownership of personal information belongs to ourselves".

What does this mean? For example, when using the Didi taxi APP, the model, time, route, cost and other information of our personal car will be uploaded to Didi's server, and massive information will form big data. The results of statistical analysis contain huge business opportunities and wealth. This statistical result belongs to Didi.

Each of us contributed a drop in the ocean to this statistical result. For example, Heping District calls an average of 500 cars a month. Living in Heping District, you contributed this data, but the information ownership of calling a car in Heping District is yours. Without your authorization, Didi Branch cannot use it.

Once my wife used my laptop to search for clothes, and the next day I went to the office to guide students. As soon as I turned on the computer, the browser pushed me a lot of lace underwear and specially marked it as something you might be interested in, which made me very embarrassed.

Merchants should not collect users' preferences and habits without users' authorization. Even if there is, it can only be used as the original data of statistical analysis and should not be used to accurately market users. This violates the principle that "personal information shall not be used without authorization" and is an abuse of artificial intelligence.

Google released the first "Made in Google" Pixel mobile phone, which uses the latest Android operating system Nougat 7. 1, and more importantly, it is equipped with Google Assistant, which is the first innovation of mobile artificial intelligence. Sundar Pichai, CEO of Google, declared that we are very happy to create a unique Google for anyone, anywhere.

However, skeptics believe that the price of this convenience is to let Google continuously obtain users' personal data and preferences, so that users can completely lose their privacy and become transparent people.

Is this artificial intelligence of Google good or bad? This cannot be judged according to personal likes and dislikes. There must be some people who like it and some who don't, so the authorization of users becomes the key. This feature should be turned off by default. When you know that your privacy will be leaked, but you still volunteer to be a transparent person to get help from artificial intelligence, you can turn on this switch.

With the vigorous development of mobile Internet, people's transparency trend is irreversible, but the right to privacy must not be completely abandoned. My philosophy is: authentication information must be absolutely independent and controllable. Is the personal preference content information open? This option should also be given to users themselves.

Artificial intelligence is in the ascendant, and the practices of various enterprises are quite different. Among the enterprises I have contacted, I agree with Huawei's approach. First of all, in terms of user authentication, Huawei saves authentication information in the form of ciphertext in the inSE of HiSilicon chip independently developed. When an Internet provider needs authentication, it compares it in the CPU and then sends the judgment result of yes orno. Because Huawei refuses to provide the user's fingerprint information to the Internet provider, there is also an industry conflict.

Some netizens may question that Huawei said that the authentication information was sealed in the chip. Is it true?/You don't say. This suspicion is quite reasonable. No one can blindly trust any enterprise by virtue of personal likes and dislikes. The essence of an enterprise is profit-seeking, and it is not worthy of absolute trust.

Then who should I trust? My suggestion is to trust the government, and you can only believe it with the endorsement of the government. To tell the truth, I didn't believe the propaganda given to me by Huawei friends at first, until I saw the Hisilicon 960 chip safety certification certificate from the central bank and UnionPay.

Recently, Huawei and Tencent had a dispute over the glory of Magic mobile phones. Everyone knows that Huawei mobile phones generally have EMUI system, but this mobile phone is equipped with Magic Live system and has some novel intelligent applications.

After the smart assistant is turned on, the system can automatically identify the keywords in SMS and WeChat, and make smart recommendations according to the location of the mobile phone. For example, when you discuss eating on WeChat, the mobile phone will recommend you a good restaurant nearby. If you don't need this function, turn off the smart assistant.

Tencent is very dissatisfied with this, thinking that Huawei stole the data in WeChat and reported it to the Ministry of Industry and Information Technology, demanding that Huawei be dealt with. Huawei's statement is that the data in the glory Magic mobile phone WeChat does not belong to Huawei or Tencent, but belongs to the user, and the user has authorized the smart assistant to use it. There's nothing wrong with that. The answer from the Ministry of Industry and Information Technology is that your two families will discuss it again.

The glory of Magic Live and Google Assistant are both innovations in mobile artificial intelligence, but they are essentially different. Google's model is "uploading personal information", and the glory model is "diving service information".

What does this mean? Simply put, Google uploads information such as users' personal preferences to the cloud. Google provides accurate artificial intelligence services after understanding user preferences.

Glory is to dive the required public service information into the mobile phone. Magic Live will break through the barriers of various apps, intelligently identify and analyze user information, and then provide accurate services.

In fact, Huawei installed a private secretary in your mobile phone. It obtains your personal information and provides accurate services, but it does not report to Huawei. It's all done inside your phone. Moreover, this function is turned off by default, and it will not work unless you authorize it to be turned on.

In addition to Google and Huawei, Apple also has similar mobile artificial intelligence services. Apple adopts the same "service information diving" mode as Glory, and does not upload users' personal information.

Why take Huawei as an example? This is because the glory Magic mobile phone opened the precedent of domestic mobile artificial intelligence, broke the barriers of various apps and re-integrated the data. This move is a major blow to the interests of app merchants, and now it is just a storm. Wrestling is still behind, which is a big event worthy of everyone's attention.

Compared with Google and Apple, Huawei's practice of sealing user authentication information in the inSE area of the chip is similar to Apple's trust area, and the diving mode of service information without uploading user information is also consistent with Apple's, while Google's practice is relatively more radical.

Artificial intelligence is a general trend, which will bring us more accurate services. However, driven by the data provided by users, it will objectively affect our personal privacy, which is a dilemma.

Mobile artificial intelligence is now in the exploration stage, and the practices of various businesses are different, and everyone's ideas are different. As an expert in the industry and an ordinary user, I have three principles for mobile phone artificial intelligence: user authentication information must be absolutely autonomous and controllable, and the decision to upload content information is up to the user himself, and no merchant can use it without the authorization of the user.