Robert Sloan, a professor at the University of Illinois at Chicago, has found an AI system to be as smart as a four-year-old. Scientists often talk about creating artificial intelligence, but how “intelligent” are these systems really?
Researchers at the University of Illinois at Chicago came up with an answer after giving one of the top artificial intelligence (AI) systems an IQ test.
The MIT-built system, dubbed ConceptNet 4, is as smart as the average four-year-old.
“We’re still very far from programs with common sense and AI that can answer comprehension questions with the skill of a child of eight,” said Sloan, who is head of computer science at the university.
His goal is research that can help focus attention on the “hard spots” or challenges in AI research.
The university reported Monday that researchers put the AI system through the verbal portions of the Weschsler Pre-school and Primary Scale of Intelligence Test, a standard IQ assessment for young children. While the system has the average IQ of a young child, unlike most humans, the machine’s scores were uneven across different parts of the test.
For instance, Sloan noted that ConceptNet 4 did very well on a vocabulary test, as well as on its ability to recognise similarities. However, the system did “dramatically worse” than average in its comprehension abilities, which are about answering “why” questions.
According to Sloan, one of the hardest problems in artificial intelligence research is building a computer program that can make good judgment calls based on any situation that might arise. Basically, it’s difficult to program common sense because scientists haven’t yet figured out how to give systems knowledge about things that humans find obvious, like the fact that ice feels cold.
“All of us know a huge number of things,” said Sloan. “As babies, we crawled around and yanked on things and learned that things fall. We yanked on other things and learned that dogs and cats don’t appreciate having their tails pulled.”