Have you ever imagined falling in love with an AI? The movie HER shows what it is like to be in love with an AI In the film, the main character got to love the female voice AI that he installed on his computer. In the movie Black Panther, AI world is described as wonderful. When an injured person lies on the AI bed, it automatically finds the wounded areas and cures them. Jarvis, the secretary for Iron Man, even recommends what to wear not to mention doing his main job as a secretary. Likewise, AI is described in many films as something intimate with humans and something that makes life more convenient. In reality, however, AI is not always welcomed. Rather, problems in various aspects are emerging.
Blurred lines of accountability in autonomous driving accidents
How much will a driver be accountable for a traffic accident that occurred while the autonomous driving system is on? The countries below stand on totally different sides. In California, a regulation that holds the manufacturer to account for the accidents was proposed. On the other hand, in Germany, the laws were amended to hold the driver for most of the accountability. The amendment is harshly criticized for its implication. It can be interpreted as drivers should account for all the accidents even the ones caused by manufacturing errors or coding errors. Germany explained that otherwise, it will make the commercialization of autonomous driving slower. The U.K. is quite neutral. Recently a bill was proposed, which says when an autonomous driving accident occurred, the assurance company should cover the expense first, and after that, the “who is accountable” procedure follows. Since regulations among countries vary, it is worried that it might cause confusion to drivers and companies.
Personal information leakage caused by face recognition
If you are a Facebook user, you probably have used face recognition function. When you upload a picture, AI recognizes the face and tells whom to tag. However, the thing is that the stored face data might be sold to other businesses or the third party. Facebook was sued by three men from Illinois where strict Biometric Information Protection Act takes effect. They claimed that the face recognition function collects bio-information without permission of its users. Also, face recognition function in smart phones is controversial. Apple introduced face ID function, which replaced existing fingerprint recognition system. This function has been under dispute. Using one’s face as a password is no different than putting a password sign on one’s forehead. If you are kidnapped, the criminals will not bother to torture you for your password. They will just put your cell phone in front of your face. The data security mentioned above is another thing that experts point out.
AI taking over human jobs
There is an expectation that AI will give humans more free time replacing humans in work fields. However, what if the replacement is forced? Imagine there is a part-time worker who works eight hours a day and is paid minimum wage. Suddenly AI stepped into his workplace and made him work one hour less. His decreased work hours will take away many portions of his wage. Decreased work hours will not make him happier. The ones who smile are the companies that hired AI They will grin with the lowered labor costs.
In extreme cases, the decrease in work hours will not be enough. Mankind could face mass unemployment. World Economic Forum anticipated that until 2020, five millions jobs will disappear with the growth of AI and the Fourth Revolution. Those who are wary of mass unemployment give an example of the Industrial Revolution. Although it took more than one hundred years, millions of people lost their jobs then. What about the Fourth Revolution? It only started in the mid-2010s but is changing many things faster than in the 19th century. AI shows outstanding ability in doing repetitive tasks regardless of the kind of labor. That is to say, AI can be a nice driver with the autonomous driving system mentioned above, and it can also be a skillful doctor. AI can utilize much data than a veteran doctor can. With that data, it will discern whether this thing is a lump or an ulcer. In fact, there already are applications that diagnose patients. It is uncertain whether the society will still need human workforce when there are cheaper a
nd more efficient alternatives waiting.
Whether you like it or not, it seems undeniable that AI will be a part of your life. To coexist with AI, humans still have a few more obstacles to deal with. The bias of AI is one of them. For instance, a negative bias of black females was detected while testing a face recognition system. A Researcher told the AI to identify 1,000 faces as a male or a female and the result was dismal. The AI gave wrong answers 34% more with dark-skinned females than light-skinned males. This problem stems from the imbalanced volume of data given in the machine-learning stage. Those who are worried about this kind of situations insist that programmers identify and correct those errors. Plus, humans have an instinctive objection to the things that resemble human on a certain level, which is called the uncanny valley. As the graph shows, when AI reaches the certain level of human likeness, the familiarity sharply drops and then goes up again. Experts expect that the improvements of the problems would help AI settle in human society.
They also say that it will definitely benefit humans in many ways when used properly. If not, it might entail various problems: the vague accountability of autonomous driving, the danger of personal information leakage of face recognition, wage cuts and mass unemployment AI might bring about. It means that AI really is a double-edged sword. Mankind is on the verge of the new age of innovation. It is unclear whether people in the future will view the advent of the age of AI positively. This is why humans have to carefully watch the use of AI and put endeavors in bringing positive results with AI. Dr. Kate Crawford, a principal researcher from Microsoft, stressed the importance of monitoring after the introduction of AI. She said, “I think it is also really important that, after releasing these sorts of complex, algorithmically-driven systems, we just continue to monitor their use across different contexts and communities.”
김보임 기자 email@example.com