Intelligence Quotient (IQ) Test is a set of standardized questions designed to evaluate human intelligence.
In June 2015, a Natural Language Processing team lead by University of Science and Technology of China and Microsoft Research gave us a surprise: They discovered a solution for machine to outperform the average human ability to answer verbal comprehension questions in IQ Tests for the first time! They documented the solution in this paper.
Verbal comprehension questions appear very frequently in IQ tests, which measure human’s verbal ability including the understanding of the words with multiple senses, the synonyms and antonyms, and the analogies among words. Here are some example verbal comprehension questions:
Isotherm is to temperature as isobar is to? (i) atmosphere, (ii) wind, (iii) pressure, (iv) latitude, (v) current.
Which is the odd one out? (i) calm, (ii) quiet, (iii) relaxed, (iv) serene, (v) unruffled.
Which word is closest to IRRATIONAL? (i) intransigent, (ii) irredeemable, (iii) unsafe, (iv) lost, (v) nonsensical.
The results of their solution are highly encouraging.
The above table demonstrates the accuracy of answering verbal questions. The RK model invented by University of Science and Technology of China and Microsoft Research can achieve the best overall accuracy than all the other methods, including the average human candidates (Bachelor and Master students). It indicates that with appropriate uses of the deep learning technologies, artificial intelligence (AI) could be a further small step closer to the human intelligence.