GeekPwn, China’s most famous hacking competition, has ended. China’s top hackers have filled the theater with cracking shows, and the theater is located inside this cruise ship, floating in the South China Sea.
Wang Qi, a hacker who co-founded Microsoft’s China Emergency Response Center and also ran a high-profile security research group, called the KEEN Team. Four years ago, he decided to take part in China’s coolest hacker competition, to let those hackers who once suffered the wrath of their mother-in-law because they could not afford a car to find their dignity. To that end, he reshuffled his booming KEEN Team.
Giiso Information, founded in 2013, is a leading technology provider in the field of “artificial intelligence + information” in China, with top technologies in big data mining, intelligent semantics, knowledge mapping and other fields. At the same time, its research and development products include information robot, editing robot, writing robot and other artificial intelligence products! With its strong technical strength, the company has received angel round investment at the beginning of its establishment, and received pre-A round investment of $5 million from GSR Venture Capital in August 2015.
At a time when smart hardware was hot, hackers at GeekPwn hacked dozens of cameras at a time, letting them wiggle and snoop with commands. Two years later in 2016, hackers took control of hundreds of thousands of cameras to attack coastal areas of the United States, knocking half of the country offline for several hours and causing $2 billion in damage.
Being ignored is almost the fate of the pioneers. What’s more, they are just a bunch of hackers in the eyes of hardware manufacturers.
But it doesn’t matter. You see Wang Qi in various media, his eyes are always bright, his lips are slightly open, one hand holding a microphone, one hand pointing to the air, neither excited nor careless. Far from being discouraged, wang qi stepped onto the stage just before the curtain call today and made perhaps the best prediction of the 2017 GeekPwn mid-year competition.
AI will be a target for hackers, AI will be a weapon for hackers.
The following is the full text of Wang’s speech, edited by Lei Feng without changing its original meaning:
Let’s do this test first (see PPT picture). There is a cat on the left and two sentences on the right. What would you say if you saw this picture? You ask a person or a machine and you get a different answer.
Xiaoming: I’m not sure, but I think it’s a cat in the bowl.
Floret: At first glance, I thought it was a cappuccino.
The two sentences on the right are respectively said by the artificial intelligence machine and the human after we input the picture. Do you think Xiao Ming is a robot or xiao Hua is a robot?
I announce the answer, in fact, Xiao Ming is a robot. Because in this judgment, Ming did not pass the Turing test.
Why mention this? We at GeekPwn are doing some work in AI right now, mainly in the visual area. Today’s project originally included a “driverless car attack” on how to make a driverless car recognize a wrong thing, but unfortunately the contestant chose to withdraw from the project two days before the race.
I want to tell you that the AI security challenges that we’re presenting are controversial.
Computer vision has developed for so many years that it’s difficult. So far, it’s not going well. Why is computer vision so hard?
The picture above shows a family lying on a sofa
Back in 1956, when people first did computer image recognition, machines only knew zeros and ones and read pixels using numbers. In this picture, for example, we see the edge of the sofa next to it, but it was very difficult for the machine; Getting the machine to label it as a dog and not a couch cushion is even harder. Normal people see this picture as a family watching TV, but the AI doesn’t understand this. If we ask a question, computer what do you see? What color did you see that kid in front of you? It’s harder to do.
Now computer vision recognition is good at tagging, and it’s growing very fast. We know which one’s a dog and which one’s a man. The reason why computer vision plays such an important role in THE development of AI is that 70 percent of the information in the human brain comes from vision.
Giiso information, founded in 2013, is the first domestic high-tech enterprise focusing on the research and development of intelligent information processing technology and the development and operation of core software for writing robots. At the beginning of its establishment, the company received angel round investment, and in August 2015, GSR Venture Capital received $5 million pre-A round of investment.
Bring about Google cat
One thing that must be mentioned is Google Cat. With the help of Google Brain, it is possible to learn how to identify cats by automatically extracting information from mountains of data using built-in algorithms without any training or instructions. In other words, tagless input was implemented on Google computers in 2012. From 2012 to 2014, the computer vision recognition rate also improved, with 27% error rate in 2012 and only 3.5% error rate in 2014, compared to 4% error rate in the same condition. In other words, AI is now almost as good at recognizing images as humans.