Abstract: Zhongzhi Shi

Title: AGI Research Progress in the Intelligence Science Laboratory at the Chinese Academy of Sciences

The goal of Artificial General Intelligence research is the creation of broad human-like intelligence, rather than narrowly “smart” systems that can operate only as tools for human operators in well-defined domains. General intelligence is the ability to solve a variety of complex problems in a real world environments. At this point in time, humans have general intelligence and computers do not. How to resolve the gap is the task of Artificial General Intelligence. And “Intelligence Science” is an interdisciplinary subject that points in the correct research direction for AGI.
Intelligence Science is dedicated to joint research on the basic theory and technology of intelligence by multiple disciplines including brain science, cognitive science, artificial intelligence and others. One of the main tasks in Intelligence Science is mind modeling, which tries to model diverse human mental activities, such as perception, learning, memory, thinking and consciousness. From the artificial intelligence perspective, the goal of mind modeling is to construct computational artifacts which combine many cognitive abilities in one integrated system, enabling artifacts that have the same intellectual capacity as humans.
A new mind model called CAM (Consciousness And Memory model) has been proposed by the Intelligence Science Laboratory at the Chinese Academy of Sciences. The CAM model consists of three parts, which are: consciousness, memory and high level cognitive function. Consciousness, as conceived here, possesses a set of planning schemes which arrange the components of CAM to accomplish different cognitive tasks. The memory part contains three types of memory which are: long term memory, short term memory and working memory. These memories are used to store different kinds of information. For example, the long term memory stores abstract knowledge like definition of concepts, actions, events etc. The short term memory stores knowledge and beliefs about the current world and the goals/sub-goals that the system intend to achieve. The working memory stores a set of information obtained from the sensors, such as visual information caught from cameras, special information caught from GPS, etc. This stored information is utilized to support the intelligent activities of CAM, e.g. visual information stored in the perceptual buffer can assist the system to detect objects in the physical world. The high level cognitive function part includes object recognition, event detection, action execution etc.
Next, the “World Wide Brain” is presented as a concrete instantiation of the CAM mind model. World Wide Brain, as we conceive it, will be a human-Internet interactive and sharing system. Through this intelligent Web, the human brain will be connected with Internet information resources, much more intensively than allowed by technologies not founded in intelligence science.
This talk highlights the CAM mind model and World Wide Brain. Directions for further research on CAM and World Wide Brain will be pointed out and discussed.

Acknowledgement: This work is supported by National Basic Research Priorities Programme (No. 2007CB311004), National Science Foundation of China (No. 61035003 , 60933004).