How do you tell just how smart your robot is? Perhaps you can give it the universal IQ test.

Traditional measures of human intelligence would often be inappropriate for systems that have senses, environments, and cognitive capacities very different from our own.



So Shane Legg and Marcus Hutter at the Swiss Institute for Artificial Intelligence in Manno-Lugano, have drafted an idea for an alternative test which will allow the intelligence of vision systems, robots, natural-language processing programs or trading agents to be compared and contrasted despite their broad and disparate functions.



Although there is no consensus on exactly what human intelligence is, most views cluster around the idea that it hinges on a general ability to achieve goals in a wide range of environments, says Legg.



The same concept can be applied to an AI system, by measuring its ability to carry out complex tasks within its particular environment and then comparing the complexity of its environment with those of a wide range of other AI systems.



“But there is a problem,” he says. Before putting this into practice the AI community will have to thrash out an agreement on what counts as the average environment. And that will not be easy.



More here.