More funding is being thrown at research to try to make computers have common sense. You’ve seen this story before, and you’ll see it again – just the players change. Related to this is the concept of the semantic web – the Holy Grail of Internet information retrieval where you will someday be able to type “Abbey Road” into a search engine and it will know that you mean the street, not the album or the studio. Or something like that!
The article makes a comparison to the sense of a child who looks out the window, sees snow falling, and knows to go put on a coat before going outside. Now, a knowledge system would be able to make this connection given this simple rule: “If snow is falling outside, then it is cold.” The snow bit flips on the cold bit. However, even after we give the machine all of these rules to work with, it still wouldn’t match the reasoning of a human who can extrapolate new rules from other related rules.
Consider if you had an erupting volcano nearby and so lava was falling. Even if you’d never seen lava before or heard of a volcano, you’d look at the lava, smell the sulfer in the air, watch the screaming villagers running away from it, and conclude that whatever this new stuff is, it must be bad news. So maybe wear two coats before going out.