Can a robot pass a university entrance exam?

25

September

2017

No ratings yet.

When we estimate how well we learn in the class, we are required to take an exam, then the score will reflect the studying results. Similarly, when we would like to estimate how smart an artificial intelligence robot can be, it is also easier and more direct to have a score. Using this score, we can also see to what extent AI robot can out-perform human being. Recently, a Japanese artificial intelligence robot team created a robot “Todai” to take the university entrance exam in Japan and compare the exam results with all the students (Todai is the abbreviation of University of Tokyo, the top university in Japan). The robot did quite well in the first stage of exams for multiple choices on different subjects and mathematics, being top 1% in the mathematics exam. However, the robot failed to answer most of the question in English test due to the reason that AI cannot understand the sentences and they cannot read. At the end, the robot did not pass the exam of University of Tokyo but it did pass the exam of other approximately 60 universities in Japan. The speechmaker pointed out that at the current stage, it was not possible to make AI robot understand the meaning of the sentence or knowledge, which is what human being can do. She encouraged educators to put more emphasis on understanding the meaning instead of only memorizing knowledge because memorization can be easily done by AI robots.

After watching this video from Tedtalk, I found it interesting because in this measurable way we are actually able to see how smart AI can be at current stage. We can also better prepare to overcome the threat AI robot could bring to the job market. However, with the high-pace development of technology and science, the possibility of AI robot being able to have motion, read and understand the meaning of sentences might not be zero. Therefore, it would be interesting to see how far the artificial intelligence can go.

 

 

Please rate this

Benefits and risks of AI

25

September

2017

No ratings yet.

In the article “The business of artificial intelligence) by Erik Brynjolfsson and Andrew Mcaffe (2017), they stated that most big opportunities of AI have not yet been tapped. I questioned myself when I read this sentence, what exactly is the big opportunities? Does it mean that when AI can do what exactly human being do, AI reaches its crest stage? And what would happen after that? Besides, should we worry about the safety of the AI technology since it has large potential? I tried searching for articles on expectations of AI technology in the future.

There is one article answers part of my concern about “safety” and “the opportunity of AI”. The article is called “Benefits & risks of Artificial Intelligence”, published by the “future of life institute”. In this article, the author discussed the potential risks of applying AI, the recent interest in AI safety, top myths about advanced AI and opinions about whether we should worry about the future AI robots/ superintelligence. Two reasons of AI being dangerous is 1) AI being programmed to do something devastating, such as military usage, and 2) developing a destructive method for achieving a goal. The example for reason 2) in the article is that self-driving car will bring a person to the airport without considering traffic rules when the person told the car to drive as fast as possible. Because of the aforementioned reasons, nowadays people are more interested in the safety problem in AI.

In addition, due to some “demonized” film or animation, people would concern about the future superman AI, who can substitute human in daily life or lead to war. However, the article stated that these misunderstanding could not be true. Currently, the superhuman AI technology has not been fully developed yet and there is still a long time to go. Furthermore, the safety concerns have been aroused, so more controls and designs would be developed. Consequently, we do not have to consider artificial intelligence as an “evil”. Being smarter and strengthening our own capability can help us to succeed in the future with the existence and competition of artificial intelligence.

For more details about the article, please find the link:  https://futureoflife.org/background/benefits-risks-of-artificial-intelligence/

Please rate this