I am often asked if I am afraid of making machines too smart with all of the work going into modeling human knowledge - the concepts, the relationships, the explicit and implicit rules.

I am not.

System checks and balances aside for now, we're not trying to create HAL.

A machine will not stop to consider random things on its own, as I did today as I was making myself a sandwich for lunch: is there a dip in pastrami sales in the summer months? I mean really, completely random. I thought it as I was trying to decide whether I should cook the pastrami or eat it cool from the fridge. I usually cook it, but it's been incredibly warm here. So I just put it on the bread. As I did so, I wondered how many others might face the same choice. And then I thought, it's really better warm. I wonder if other people like it better warm? If they do, and don't want to eat it cool, do they buy less of it in the summer months? I think overall we do. Hmm. And then I proceeded to sit and eat.

Even if we detailed of my own personal connections and 'rules' about life, a computer would never randomly ask itself about pastrami. Ever. That is still reserved for the machinery of the human brain.

If some day in the future a computer does ask itself about pastrami, then is it still a machine? We'll have to dig up Gene Rodenberry's thoughts on the matter in countless Star Trek tomes and contend with greater societal issues.

Either way, it's all good. And exciting. And not fearful.