This paper is the extension of my recent paper which was presented at the AGI-22 conference. In this paper, I try to answer the comments I received during and after the conference and to clarify and explain in more details the points and results that were missed or omitted from my previous paper due to the page limitation of the proceedings.

In this paper, an index for measuring the versatility and efficiency of artificial general intelligence (AGI) systems is proposed. The Versatility-Efficiency Index (VEI), is the updated version of our previous efforts (i.e., Versatility Index (VI)) towards a comprehensive definition of an intelligence quotient (IQ) for intelligent agents. VEI is based on both Legg-Hutter and Pennachin-Goertzel definitions of intelligence and plays as an alternative way for measuring the intelligence level of intelligent agents. VEI, in contrast to VI, also encompasses the qualitative characteristics of intelligent agents like their wellness of performance and the complexity of the operating environments. VEI is applicable to both of the natural general intelligence (NGI) agents and AGI agents. For determining two parameters of VEI, AGI Pyramid – a novel classification of environments by classification of the problems of the universal problem space (UPS)- is proposed. The role of the Artificial General Intelligence Society (AGIS) in the mentioned classification and determination as well as the importance of the VEI in slowing down or preventing from singularity and its role as the possible bridge between intelligence and physics are also discussed.

In this paper, an index for measuring the versatility of artificial general intelligence (AGI) systems is proposed. The index called Versatility Index (VI) is used to measure the versatility of an AGI system or for comparison purposes between different AGI systems. Then, an upgraded version of the original AGI Brain is proposed. In the new model, AGI Brain II, the explicit memory (EM) is replaced with a modified Mamdani fuzzy inference associative memory, called ProMem, which is able to estimate the consequences of a certain action by estimating the probability density function (PDF) of the observed data in a stochastic environment. The model was tested in a portfolio optimization scenario as a stochastic environment. Simulation results demonstrate the accuracy of the novel explicit memory as well as the increased versatility index of the upgraded model.

In this paper a unified learning and decision making framework for artificial general intelligence (AGI) based on modern control theory is presented. The framework, called AGI Brain, considers intelligence as a form of optimality and tries to duplicate intelligence using a unified strategy. AGI Brain benefits from powerful modelling capability of state-space representation, as well as ultimate learning ability of the neural networks. The model emulates three learning stages of human being for learning its surrounding world. The model was tested on three different continuous and hybrid (continuous and discrete) Action/State/Output/Reward (ASOR) space scenarios in deterministic single-agent/multi-agent worlds. Successful simulation results demonstrate the multi-purpose applicability of AGI Brain in deterministic worlds.

In this paper a novel computational behavior model is proposed which has a simple structure and also includes some of the major affecting parameters to the decision making process such as the agent’s emotions, personality, intelligence level and physical situation. The effect of these parameters has been studied and the model has been simulated in a goal-achieving scenario for four agents with different characteristics. Simulation results show that the behavior of these intelligent agents are natural and believable and suggest that this model can be used as the decision making and behavior control unit of future life-like intelligent agents.

Modeling behavior of intelligent agents and its affecting parameters is a very challenging aspect of research in the field of Artificial Intelligence. But, if performed correctly, we can improve the abilities of artificial agents and we can build social agents which can speak, think and behave like us. Many other models of behavior for intelligent agents have been proposed but their complexity makes it difficult to validate them against the real human decisions. In this paper a novel behavioral model is proposed which has a simple structure and also includes the effect of emotions as a major affecting parameter to the decision making process.