Gigaom brings you our unique analysis and commentary on the present and future of AI.
I'd like to spend the next few episodes of the AI Minute talking about common sense. Common sense, it turns out, is a thing that we don't know how to teach computers and it's really at the core of our intelligence. Now, you know we say sometimes the so-and-so is smart, but they don't have any common sense and that's a little bit what I'm talking about but I mean something more basic.
A dictionary definition of common sense is something like "sound judgement from a simple perception of facts." It's what obviously follows from facts. And by that I mean things like you don't walk off a cliff, things that a human doesn't even have to think twice about. And yet, a computer has a hard time understanding why you wouldn't walk off a cliff.
If you start to deconstruct it, you realize that at its core there are things about goals and consequences, and cause and effect, and good outcomes and bad outcomes, and all of these other factors that somehow never enter our calculus when we're deciding not to walk off a cliff. In fact, we didn't even consider walking off a cliff. So how do you capture all of that stuff that people know almost intuitively, but not quite and teach it to machines? That's a really big problem in artificial intelligence.