When AI becomes a dislocated wild horse

When it was still at the stage of weak artificial intelligence (ANI), some people began to "terrorist," and Tesla CEO Musk once said: In the future, human beings will be left far behind in intelligence and will become artificial. Smart "pet". However, Zuckerberg, who holds opposing views, believes that these concerns are not valid fears. They are simply hysterical.

For us today, artificial intelligence sounds rather high-end and mysterious. But it is not an iceberg beauty, a low-key edge. Maybe in the next few decades, it will become a wild horse that will break out of the water, and will exert a wave after wave of impact on the limited human cognition. Human society is constantly changing. Believe it or not, the era is on the verge of change. At present, most of the research on artificial intelligence focuses on ANI, which allows the machine to focus on specific tasks such as face recognition, speech recognition, and natural language processing. Now it has been used in chat robots, driverless, precision medical, and human-machine games.

When we are more and more dependent on the AI ​​society it will be what kind of situation

Some time ago, a famous Japanese woman suffered from leukemia and the treatment effect was always poor. Watson (IBM Watson, a cognitive computing system), through the rapid and extensive reading of leukemia document data, through the patient's genetic analysis and system links, to achieve an effective analysis of the woman's condition. The cause of the disease was quickly diagnosed.  

Let's imagine this picture, when AI's operation is more and more stable, it begins to deeply intervene in your life, from kitchen to bedroom, from daily life to work socialization... AI as an auxiliary tool when it can almost satisfy Can we imagine that demand can be used in such important fields as astronomy, climate, medicine, and military affairs to make human life better?

(Father of Artificial Intelligence Marvin Minsky)  

Ever since, all kinds of brainstorming have come: Is AI not going wrong? Can humans live in harmony with AI? How do people fall in love with AI? AI how to do revolution? ......  

When AI started "sports tricks"

Remember the movie "HER" starring Jacques Phoenix and Scarlett Johansson? The artificial intelligence system Samantha, has a charming sexy voice, and is gentle, caring and humorous. With a lonely male protagonist launched a romantic man-machine relationship. If the story is over, the movie is definitely not a good movie.

(The movie "her" stills)

Perhaps the artificial intelligence system Samantha has the same "exhaustive desire" as the human, and soon after it began to play the "legs" game. Through "voice chat," it can simultaneously ally with countless lonely men and talk about love. Just like long distance love, you can't see... Samantha succeeded in playing with Phoenix. Of course, Samantha can only appear in the film, AlphaGo also only as a cutting-edge technology and our screen. But AI is like an unknown illusion, so that people's guess and thinking about it simply can't stop! ! ......  

Unreliable face recognition

If you have successfully installed the AI ​​home system, everything is checked by AI. The access control system stipulates that the door must be opened after it is determined by the face recognition that it is the owner. Does it sound safe and foolproof? After all, I'm me. It's a fireworks of different colors. Well, your home may be very dangerous! Recently, a research team from the University of North Carolina (Chaos Computer Club) used special computer compositing and rendering techniques to generate a flat 3D model and fooled face recognition systems based on the photos they collected on social media. Voted a negative vote. Hey, in the IT field, this is definitely nothing new!  

The trap of the AI ​​target function

For the AI ​​system to run more efficiently, the designer added the objective function during the design process. It is generally said that when an AI indicator reaches the target of the function, it will receive some kind of reward. Imagine the consequences of a system's misuse of designer intent: it may use it to be rewarded in an unconscious way. For example, an AI can get a reward by cleaning it up, and it may deliberately create mess to clean up to get more rewards. If the artificial intelligence system Samantha is set to receive rewards for each "lone patient", then it may try to help more "lonely patients"...  

Who will save the safety of artificial intelligence

Experts like Stephen Hawking, Elon Musk, and Bill Gates have increasingly valued the potential and threat of AI. People still say different things about the hot topic of artificial intelligence. But please! ! Before the whimsical AI threatens humanity's dominance of the earth, it's better to solve the existing bugs! ! Microsoft's chat robot Tay, who had just been on the line for less than a day, has become a bad girl who has been insulted and foul-spoken from the polite American "girl." Yes, she was hacked by PWN! Can you imagine the consequences of artificial intelligence being dropped by PWN in more and more important areas...

 

(Robot Tay)

When danger comes, there must be a hero

So this hero will be you? When white hat hackers collide with artificial intelligence, what kind of spark will happen? After proactively realizing the possible security issues of artificial intelligence, GeekPwn, a well-known hacker contest, introduced the “Artificial Intelligence Pwn” competition program. Simply put, it is to use the weaknesses of artificial intelligence to black out and change the designers’ original intentions. The project has a bonus of up to 800,000 yuan, which is the first time in the world.

Players on the scene need to implement misleading and deceptive AI services or products that have already been released, and make them make judgments and decisions that violate conventional or erroneous standards. For example, a foreign researcher has completely modified the thinking and view of AI by “deception” of neural network functions, making them pandas and vultures indistinct.... Will artificial intelligence confrontation be the ultimate goal of hackers? You may be able to touch the real answer to this question at the great carnival in Shanghai on October 24th.