Task Accomplishment to Verify True Intelligence in AI Agents

Abstract

The concept of artificial intelligence was introduction to the world by Alan Turing in 1950. From then many researches are done to strengthen the AI. By developing new techniques in this field scientists are trying to make robots that are efficient in various domains of intelligence such as perception, reasoning, decision making, self learning, natural language processing, problem solving etc. While doing this, researchers have come across various problem and difficulties and are engaged in solving those. Thus many a time research, problem and future regarding the AI are mostly discussed. Even the negative aspects of AI are also seen being discussed so as to avoid any further complication in future but rarely is there any discussion on how to test whether the so called intelligent robots are really so? Alan Turing, the father of AI, himself tried to answer this question by suggesting the famous ‘Turning Test’. But its inadequacy was proved by another famous test called the ‘Chinese Room Test’. The problem is the most intelligent looking acts such as conversing and playing chess are in reality just an illusion caused due to smart computer techniques or advanced mathematics. These are not based on the true ability to think. So considering this theme as our focal point we have tried to explore a few ‘tasks’ when provided to an ‘intelligent’ machine will prove its degree of intelligence. We have used various domains of intelligence to formulate a task. This paper will explain about how to differentiate between true intelligence and the imitation of intelligence with the help of these tasks. Accomplishment of these tasks by any virtual machine will open many new possible application of AI and also these tasks may prove as guideline for further research techniques.Also we have tried to justify the reason to continue the research and development of AI for betterment of humankind.