Artificial intelligence (AI) is already a very popular topic, so much so that you may mention it consciously or unconsciously when you chat with your friends. The prospects are fascinating, and we keep hearing amazing stories about efficiency, automation and intelligent prediction.
But artificial intelligence is not perfect. Behind every success story, there may be many failed “mothers”. Things did not go according to plan. Although I am a loyal believer of artificial intelligence and have seen its power in many businesses, it is a good thing to look at things dialectically and objectively.
The intelligent dialogue robot has learned some inappropriate languages
Nowadays, artificial intelligence dialogue robots have become the norm on social media and other websites to some extent. But it is not perfect. Microsoft’s artificial intelligence dialogue robot is an example, which is referred to as “Tay” for short.
Microsoft’s vice president of artificial intelligence and research, Peter Lee, had to apologize publicly for not foreseeing this possibility in advance.
Amazon has a project called Rekognition, which is a facial recognition software based on artificial intelligence. When sold to law enforcement agencies, it is mainly used to investigate crimes. It essentially cross-analyzes the images and guides law enforcement officers to find possible suspects.
The problem is that it is not very accurate. In a study, dozens of photos of athletes in the Boston area were retrieved in the system. Among these athletes, at least 27 people-about one in six-were incorrectly matched with the criminal’s facial photos. Among them is a rugby star who has won the Super Bowl three times.
Apple is always introducing cutting-edge technology. Over the years, they have been setting standards for the smartphone and mobile device industries. In most cases, these products are enthusiastically pursued by everyone. But maybe it is too eager to chase, and sometimes it will “turn over.”
Before the release of iPhone X, Apple invested a lot of time and money in their facial recognition system, which replaced the fingerprint reader and became the main method of accessing mobile phones. They claim that artificial intelligence components are very smart, and readers can wear glasses, make-up, etc., without affecting functionality. This is basically correct. The problem is that Apple also made it clear that Face ID technology will not be deceived by masks or other technologies.
A Vietnamese security company sees this as a challenge. For only $200, they made a mask out of stone powder, glued some 2D printed “eyes”, and unlocked a mobile phone.
Who doesn’t like the idea of robotic dogs? You get a cute little machine that does not bark, does not need to eat, does not urinate and urinate anywhere, and does not incur expensive veterinary bills.
ston Robotics was shown at a conference in Las Vegas, it “died” dramatically prematurely on stage. Its task was to walk, but slowly began to stagger and eventually fell to the floor.
IBM’s Watson is an AI platform for enterprises and has achieved many achievements.
In 2018, Watson tried to launch a medical artificial intelligence system to provide advice for treating cancer patients. But it didn’t take long for the hospital and oncologists to discover the major flaws. Watson once suggested that patients with excessive bleeding should take medication, which may cause more bleeding-and even cause the patient to die during the medication!
The IBM investigation found that engineers used hypothetical and fictitious cases for artificial intelligence systems, rather than relying on actual patient data and historical medical records.
Deepfakes is the most controversial AI software. It is becoming a serious (and worrying) problem. Hackers have found ways to fake sounds, pictures and even videos. In some cases, the impact is catastrophic.
In March 2019, the CEO of a British company received a call from the boss of the German parent company. He was ordered to transfer funds equivalent to US$243,000 to a Hungarian supplier. The request was marked as urgent and was told to execute it immediately. As a result, everyone must have guessed that it was not his boss on the other end of the phone. Use an artificial intelligence-based software to imitate the boss’s voice.
Of course, in a sense, the artificial intelligence software has won, and humans have been defrauded of more than two hundred thousand dollars!
Artificial intelligence is not perfect. So, as we look forward to the next three or five years, let us remember that artificial intelligence is still developing.
In order to achieve the positive effect we want, we must deal with many problems and obstacles in the process of using AI. In fact, most of the products mentioned in the “rollover accident” today have been greatly improved. How we deal with these anomalies and defects ultimately determines our future development direction.