Go to contents

THE DONG-A ILBO Logo

The fourth principle of AI

Posted August. 02, 2017 07:33,   

Updated August. 02, 2017 08:07

한국어

A video clip that two Google Homes (AI speakers) were talking to each other was posted on YouTube early this year and surprised many people as AI speakers were considered to talk to humans only. Their conversation was all mixed up. For example, when a Home asked ‘What is the love’, the other one answered that ‘Baby don’t hurt me.’ However, it was somewhat shocking to me as we do not know how things are going to end up. If they continue to talk to each other, they could get out of human control someday.  

 

It is true that research institutes studying on AI make their AI gadgets talk to each other. Researchers do that to check the voice recognition function to see whether their AI devices clearly speak and understand meanings. Such approaches are gradually moving to a direction to test an ability to deliver and accept knowledge with speaking. For instance, an AI with an outstanding knowledge about movie talks to other AI that knows the music well and they continue an interaction learning.

Two chatbots of Facebook AI Research (FAIR) accidently created its own language and developers forcibly shut down the system. Chatbot Bob said, “I can I everything else,” and Chatbot Alice said, “Balls have a ball to me,” and repeated ‘to me’ for seven times. Of course, their conversation does not make any sense, but it seems shorthand expressions. It is like for the teens to talk to each other by using their own abbreviations that adults cannot understand.

Russian-born, American author Isaac Asimov announced the three laws of robotics in his 1942 novel titled "Runaround." Firstly, a robot may not injure a human being or, through inaction, allow a human being to come to harm. Secondly, a robot must obey the orders given it by human beings except where such orders would conflict with the First Law. Thirdly, a robot must protect its own existence as long as such protection does not conflict with the First or Second Laws. Probably, we might add the fourth principle, which reads, a robot must speak with a normal language that humans can understand.