Search Engine IQ Test Based On The Internet IQ Evaluation Algorithms

The development of Artificial Intelligence (AI) in the future may have a potential threat on human, and now, more and more scientists and entrepreneurs have expressed their fears and concerns. Such potential threat has sparked a huge controversy in the society, behind which, is essentially the issue whether the artificial intelligence system can quantitatively evaluate its intellectual abilities. With the issue as the starting point and basis for the research, this paper analyzes the difficulties that the artificial intelligence ability evaluation is facing, and identifies that the intelligence concept is not well defined; the evaluation methods for the human intelligence and the artificial intelligence system are not uniform; the limitation in the technology of the artificial intelligence system itself and many other factors are the keys contributing to the problem mentioned above. Previous scientists have made a lot of meaningful work for the evaluation system of the artificial intelligence, but the quantitative analysis on the intelligent ability of the artificial intelligence system has always been a bottleneck problem not resolved yet. Under the context of the rise of big data and the rapid development of the Internet artificial intelligence, the following work is carried out in this paper in terms of the issue of the artificial intelligence system quantitative evaluation and the innovations described below are generated:(1) Based on the deep combination of the artificial intelligence with the big data of the Internet and the studies on "Internet virtual brain" architecture generated by cross-over comparison of the Internet and the brain science, the issue of the continuous intelligence evolvement in the Internet and its applications is proposed. Thereby, the IQ-related concepts like the Internet intelligence and the Internet applications are generated and defined.(2) To solve the problem of the lack of a unified test for both the artificial intelligence and the human intelligence, a "standard intelligent system model" is established on the basis of predecessors’ work, and the characteristics of the artificial intelligence system and life systems such as human life system are defined. Besides, the" Standard Mathematical Model for Intelligent Machine"、"interactive model of standard intelligent system data, information and knowledge"、"standard intelligent system evaluation model" and "Extended Von Neumann architecture" are also set up in this paper.(3) The Internet IQ scale and the evaluation methods are developed based on the standard feature model of the intelligence system, and the intelligence evaluation system suitable for both search engines and human, namely - the InternetIQTest (B/S mode) system is developed.(4) The InternetIQTest system (supported by manual testing) is applied to test 50 search engines in the world and three different age groups of human subjects, and a unified ranking list of Internet absolute IQ and deviation IQ is generated. At the end of the paper, the K-means clustering algorithm is applied in the clustering analysis, and the effectiveness of the clustering algorithm is verified via the support vector machine algorithm (S VM) and other algorithms; The method of stratified sampling is applied in Verification of the stability of test question database. According to the test results, the weaknesses of the artificial intelligence in comparison with the human intelligence are analyzed and proposals in terms of further development of search engines and relevant systems are presented as well. Meanwhile, it is also pointed out that focusing on the development of the "knowledge innovation capability" is helpful to greatly improve the intelligence level of search engines.