People who enjoy technology usually have the same dream, to have their devices able to do some of the most intricate features while still being able to interact with people as casually as a normal human being. From the responsive computer on Star Trek to Tony Starks faithful AI companion in "Ironman," JARVIS, people have been hoping for an artificial intelligence to help them with every day tasks for quite some time. Now, thanks to some new innovation and creative understanding of the human brain, scientists are closer to that technology than ever before.
ConceptNet 4 is an AI developed at MIT that is less concerned with having the most processing power and more concerned with building and intelligence that will make connections similar to a practical human brain. The system is a semantic network containing a large store of information that can be used to teach the system in terms of concepts rather than tasks, according to ExtremeTech. As the website puts it, the ConceptNet would understand that a saxophone is a musical instrument in the same way any other computer or AI would "know" that information. What makes ConceptNet unique is its ability to process the relationship between a saxophone and other things. It would be able to understand that the saxophone is used extensively in jazz music.
In order to test the scope of this AI, the creators recently put it through a standard IQ test given to young children, which yielded some strange results. Researchers at the University of Illinois at Chicago decided to take the AI and see how it faired on the Wechsler Preschool and Primary Scale of Intelligence Test. The AI passed the test getting marks that were on par with a four-year-old child's average intelligence... almost.
The team points out that it would be concerning to find a real child whose results were as lopsided as those of the ConceptNet. The system performed above average on parts of the test that have to do with vocabulary and recognizing the similarities between two items, mainly because that's how the AI is designed. For example, a typical question from this area might be "What do apples and bananas have in common?"
Where the AI didn't do as well was with questions that had more to do with comprehension, which test one's ability to understand practical concepts. For example, a question from this section might have read: "Why do we shake hands?"
The conclusion to take away from this test is that there's something to the idea of introducing intelligence to computers as a means of studying concepts. However, as the IQ test ConceptNet 4 took shows, people are very far off from having an artificial productivity companion that can do more than a slightly odd four-year-old child.
