Slouching Toward Singularity –“Artificial Intelligence System Tests Equal to a Human 4-Year Old”





Artificial intelligence will surpass human intelligence after 2020, predicted Vernor Vinge, a pioneer in AI, who warned about the risks and opportunities that an electronic super-intelligence would offer to mankind.

Vinge was a professor of mathematics, computer scientist, and science fiction author who is well-known for his 1993 manifesto, "The Coming Technological Singularity," in which he argues that exponential growth in technology means a point will be reached where the consequences are unknown. He is best known for his Hugo Award-winning novels A Fire Upon the Deep, in which he envisions a galaxy that is divided up into 'zones of thought', in which the further one moves from the center of the galaxy, the higher the level of technology one can achieve.

"It seems plausible that with technology we can, in the fairly near future," Vinge continued, create (or become) creatures who surpass humans in every intellectual and creative dimension. Events beyond such an event — such a singularity — are as unimaginable to us as opera is to a flatworm."

Now, MIT Technology Review described the findings to the question to what extent do AI capabilities add up to the equivalent of human intelligence? The results have been released on ARVIX by a team from the University of Illinois at Chicago and an AI research group in Hungary. The team compared an AI system, ConceptNet, an open-source project run by the MIT Common Sense Computing Initiative, to a standard IQ test given to humans.

The team administered the Verbal IQ (VIQ) part of the Wechsler Preschool and Primary Scale of Intelligence (WPPSI-III) to the ConceptNet 4 AI system. The test questions (for example, “Where can you find a penguin?” and "Why do we shake hands?") were translated into ConceptNet 4 inputs using a combination of the simple natural language processing tools that come with ConceptNet together with short Python programs. The question answering used a version of ConceptNet based on spectral methods.

The ConceptNet system scored a WPPSI-III VIQ that is average for a four-year-old child, but below average for 5 to 7 year-olds. Large variations among subtests indicate potential areas of improvement. In particular, results were strongest for the Vocabulary and Similarities subtests, intermediate for the Information subtest, and lowest for the Comprehension and Word Reasoning subtests.

Comprehension is the subtest most strongly associated with common sense. The large variations among subtests and ordinary common sense strongly suggest that the WPPSI-III VIQ results do not show that "ConceptNet has the verbal abilities a four-year-old." Rather, children's IQ tests offer one objective metric for the evaluation and comparison of AI systems.

The Daily Galaxy via ARVIX and MIT Technology Review

Image credit:

"The Galaxy" in Your Inbox, Free, Daily