The existing AI systems are as smart as a 4-year-old child, says a new research that entailed comparing the former with the human IQ. The findings are published online at arXiv.org.
The scientists conducted a series of tests to evaluate the smartness of the AI systems. Our best ones are only equivalent to toddlers in terms of intelligence level. Furthermore, when pitted against seven-year-old intelligence, the programs lagged behind significantly.
While computers are better than humans – regardless of the latter’s age – when it comes to the speed of calculations and similar aspects, software cannot match the human ability to have a rational understanding of the environment it is in.
One of the systems tested was the AI machine developed by MIT called the ConceptNet. It displayed high scores for vocabulary and similarities, did fairly well on information, and only poorly on word reasoning and comprehension.
The IQ tests used were of a lower level (they normally differ in scope and format, depending on age), and questions such as “what is a house?” and “where can you find a penguin?” were involved. Computers obviously cannot relate to this kind of questions. When the ConceptNet system was presented with a set of clues going along the lines of “this animal has a mane if it is male”, “this is an animal that lives in Africa” and “this a big yellowish-brown cat”, it answered with a list of potential matches: dog, farm, creature, home, and cat.
“Common sense should at the very least confine the answer to animals, and should also make the simple inference that if the clues say it is a cat, then types of cats are the only alternatives to be considered,” said one of the researchers, Stellan Ohlsson, in a statement to MIT Technology Review.
Well, it seems that we are very far away from the age of Ultron.