Glitch Deep pupil of CTO Zhao: We should be proud to call AI pseudo intelligent

Absrtact: The previous stage Future forum opens in Beijing, the biologist Rao is not polite to call the present artificial intelligence the pseudo intelligence. This afternoon 36 Krypton held an online salon activity on the subject of AI, and Glitch's Zhao, the CTO of the Deep pupil, also

The former stage of the Future Forum opened in Beijing, biologists Rao not polite to the current "artificial intelligence" known as "pseudo intelligence." This afternoon 36 Krypton held an online salon activity on the subject of AI, and Glitch, the CTO of Zhao, expressed a similar view in the event. Interestingly, he thinks, we should be proud to call AI Pseudo intelligence.

Zhao mentioned a story: the domestic before a group of criminal groups focused on ATM fraud, they greatly and small move more than 50 machines, through the ATM machine camouflage sockets, installation pinhole camera to steal credit card account password. Although there are many surveillance cameras on the banks and ATM machines, the effect of monitoring is compromised because there are too many monitoring devices for security personnel to see. Similar monitoring of these things can be done through computer vision using machines, which will benefit mankind a lot.

Another example is car driving. Since the day the car was invented, the total number of people who have killed it has been more than all the wars of mankind have combined. This is a very scary number. The problem with driving in a car is that everyone has a limited amount of time to focus. Automatic driving may solve this problem, and to achieve this step, first of all need to rely on computer vision, let the car itself "understand" the driving environment, and respond.

The two examples give me the feeling that machines are sometimes better suited to do certain things than people, even though people are much smarter than machines. The problem now is that machines lack an "entry" or an "input" capability in dealing with this kind of business that consumes a lot of human energy: in the case of monitoring, "depth recognition" is such an input capability. Two of people staring at a bunch of surveillance tapes is certainly less efficient than machine handling, but the machine cannot find a suitable "portal" if it does not have the ability to "recognize" the image correctly before processing the video. If the "portal" for dealing with such matters is resolved, or if the entry threshold is lowered, the utility of the machine will be very substantial.

and lowering this threshold is the real meaning of artificial intelligence.

The machine's vocation is in some ways better for people to provide services, artificial intelligence to solve is to provide more services to the machine portal. Because of this, people should embrace the name of AI Pseudo intelligence in a proud gesture.

Unfortunately, people's understanding of artificial intelligence has always been biased, too many people to simulate, or even breakthrough the wisdom of the human Brain Project in artificial intelligence, and thus produced a pile of "singularity" issues of discussion. But know that the human brain has about tens of billions of neurons, the magnitude of which is similar to the number of stars in the Milky Way. The exploration of the universe so far is still a very primitive early stage, especially for the brain. The day is far away from us to allow AI to reach the level of intelligence that reproduces the brain. For the singularity, we temporarily hold the posture of "looking up" is good.