Gary Marcus from ‘The New Yorker’ started his talk about ‘Artificial Intelligence’ raising eyebrows saying that even Mr. Elon Musk is worried about the future of AI. Fifty years after creation of computer program for playing Chess, computers can today beat any human being in this game. The development of algorithms and the ability of computers to gradually learn from repetitive task shows that there is a gradual growth in AI. The only question that keeps technologists up at night is that when will AI become so intelligent that they will solve our everyday problems.
When experts were asked over 50 years ago about AI becoming a reality for solving everyday problems, they said that they were just 20 years away. However, Experts have been saying this for last 50 years. In 1973 when a report was released that AI would not be able to scale up the performance of Britain, funding for AI dramatically decreased almost killing AI during this time. Today, AI has enabled robots to perform operations of driverless cars, language translation, etc. These are very special purpose and narrow robotic applications using AI but we have still not been able to develop generalized applications using AI which needs some common sense such as how to take care of children, how to cook food, etc.
IBMs Watson computer has been able to beat human beings in Jeopardy. The ability of Watson to look for keyword searches from Wikipedia pages has enabled it to beat humans. However, 3 years after beating human beings in Jeopardy, IBMs Watson has not been able to achieve anything else. This shows that AI is still not progressing above its specialized applications. Google Inc. is also focusing on the growth of AI. In that regard, Google acquired an AI company like DeepMind- founded by Demis Hassabis, a 37-year-old neuroscientist and former teenage chess prodigy, along with Shane Legg and Mustafa Suleyman.
Today’s AI technology is far away from comprehending human questions in order to offer proper and logical answers to our questions such as
Q1. How much do transistors cost in Intel Pentium?
Q2. What medicine is useful to avoid specific health problems ? etc.
This shows that there is a huge market for growth of AI. If world economy is close to 7 trillion then AI based smart machines can easily earn up to 1 Trillion. There are two reasons that we are not able to achieve it. One reason being that AI has fallen in love with statistics and second AI has also become dependent on Big Data. Google Translate is one such application of Big Data. Big Data collects a lot of data and does correlation just like children ask questions. While AI technology uses a lot of data in its analysis, the children make use of small data and gradually learn from it. Additionally, Machines do not possess common sense which human beings possess. It will be possible for AI to try to get to that level only if there is a better understanding of human intelligence.