Ever since computers were first invented, scientists and futurists have dreamed of the day when computers might be capable of autonomous reasoning and be able to surpass human beings. In the past few decades, it has become apparent that simply throwing more processing power at the problem of true artificial intelligence isn’t enough. The human brain remains several orders more complex than the typical AI, but researchers are getting closer.
One such effort is ConceptNet 4, a semantic network being developed by MIT. This AI system contains a large store of information that is used to teach the system about various concepts. But more importantly, it is designed to process the relationship between things. Much like the Google Neural Net, it is designed to learn and grow to the point that it will be able to reason autonomously.
Recently, researchers at the University of Illinois at Chicago decided to put the ConceptNet through an IQ test. To do this, they used the Wechsler Preschool and Primary Scale of Intelligence Test, which is one of the common assessments used on small children. ConceptNet passed the test, scoring on par with a four-year-old in overall IQ. However, the team points out it would be worrisome to find a real child with lopsided scores like those received by the AI.
The system performed above average on parts of the test that have to do with vocabulary and recognizing the similarities between two items. However, the computer did significantly worse on the comprehension questions, which test a little one’s ability to understand practical concepts based on learned information. In short, the computer showed relational reasoning, but was lacking in common sense.
This is the missing piece of the puzzle for ConceptNet and those like it. An artificial intelligence like this one might have access to a lot of data, but it can’t draw on it to make rational judgements. ConceptNet might know that water freezes at 32 degrees, but it doesn’t know how to get from that concept to the idea that ice is cold. This is basically common sense — humans (even children) have it and computers don’t.
There’s no easy way to fabricate implicit information and common sense into an AI system and so far, no known machine has shown the ability. Even IBM’s Watson trivia computer isn’t capable of showing basic common sense, and though multiple solutions have been proposed – from neuromorphic chips to biomimetic circuitry – nothing is bearing fruit just yet.
But of course, the MIT research team is already hard at work on ConceptNet 5, a more sophisticated neural net computer that is open source and available on GitHub. But for the time being, its clear that a machine will be restricted to processing information and incapable of making basic decisions. Good thing too! The sooner they can think for themselves, the sooner they can decide we’re in their way!
One thought on “Judgement Day Update: A.I. Equivalent to Four Year Old Mind”
Not unless they’re like K-9 from Doctor Who. That little guy would never hurt anyone…unless they tried to harm the Doctor or Sarah Jane Smith. Then he’d start using laser beams.