ARTIFICIAL INTELLIENCE QUOTIENT (AIQ)

THIS PROJECT IS INTENDED TO IDENTIFY MEANS OF EVALUATING ARTIFICIAL INTELLIGENCE SYSTEMS BY CONTRASTING THEIR ACTUAL LEVEL OF PERFORMANCE TO THE LEVEL THAT WOULD HAVE BEEN EXPECTED FOR THEM. IN SOME WAYS, THE CONCEPT IS RELATED TO THE INTELLIGENCE QUOTIENT, WHICH IS A RATIO OF MENTAL AGE TO CHRONOLOGICAL AGE, A COMPARISON OF THE PERFORMANCE NORMS TO THE INDIVIDUAL PERFORMANCE LEVEL. FOLLOWING THIS CONCEPT, THE AIQ WOULD COMPARE THE PERFORMANCE OF AN AI SYSTEM WITH PERFORMANCE NORMS. THE CRITICAL TASKS ARE TO SELECT A RANGE OF PERFORMANCE EVALUATION TESTS AND TO DEVELOP A MEANS OF GENERATING NORMS FOR A SPECIFIC AI SYSTEM. THE APPROACH SELECTED WILL RELY ON COMPARISONBASED PREDICTION METHODS TO ESTABLISH THE PERFORMANCE NORMS AND WILL SELECT TASK ANALYTIC PROCEDURES FOR SYNTHESIZING A PERFORMANCE EVALUATION TEST BATTERY FOR ANY SPECIFIC AI DOMAIN. SUCCESSFUL DEMONSTRATION OF THE FEASIBILITY OF THESE METHODS WILL LEAD TO A PHASE II DEVELOPMENT OF A GENERALIZED PERFORMANCE TEST BATTERY, GENERALIZED PERFORMANCE NORMS, AND THE ESTABLISHMENT OF A DATA BASE TO UPDATE PERFORMANCE NORMS TO REFLECT TECHNOLOGICAL PROGRESS. THE PRODUCTS OF THIS RESEARCH WOULD BE THE GENERATION OF AN AIQ METHODOLOGY AND THE COLLECTION OF DATE TO MAKE THE METHODOLOGY WORK.