Gigaom brings you our unique analysis and commentary on the present and future of AI.
Machines can store massive amounts of data perfectly, but we cannot. But we make up for this with a clever trick; we store metadata. We store conclusions about things. We save the conclusions about multiple experiences we've had throughout life.
For instance, have you ever had a conversation where somebody says, "You always do ____ and ____"? And you say, "No, I don't. Name one time I did that." And then they're like, "Well, I can't name one time, but I just know you always do it." And that's a case where you save a conclusion and throw away the individual facts.
Then we do something else that's kind of interesting. We take all of the conclusions and we can generalize them across multiple domains. So if I were to ask you, what are all the ways that a phone pole is like a tree, you could answer that. And how is a tree like a daffodil? We can answer that. Our minds are collections of interrelated metadata, and it is the whole of this, its entirety, by which we achieve general intelligence. Can a computer do this? It is unproven, but perhaps possible.