descartes does AI
… in the sense that my semi-proof here follows descartes’ formula of stripping everything away and then putting it back.
Pertaining to the “machines can only do what they’re programmed to do, and never think or feel or anything” debate….
I had a point in the debate the other day that I’ve decided I want to try to expand further. Actually, I don’t remember if we used it in the debate or not, but the Turing side (go team) discussed it prior to the debate anyway. The issue is where to draw the line between thinking and deciding. If I’m standing in a buffet line or ordering off a menu somewhere—both situations hypothetically where it would take the exact same amount of time, effort, or money to get either option being contemplated—and I don’t really have anything in mind but I still have to make a decision, isn’t that thinking? We call virtually anything we do in our heads “thinking,” be it deciding, reasoning, loving, raging, whatever. So if a computer has a choice to make where either option is equally beneficial and equally hazardous, isn’t that exactly analogous to me deciding what to eat? And if it’s exactly analogous, doesn’t that mean machines can think?
Ah, I hear the arguments already, just because machines can think doesn’t mean they can emote. That’s the central point to the “deciding” argument is that there are no emotions involved. But, let’s take a nearly Cartesian leap here (in the sense that the logic may not necessarily follow) and say that there are emotions involved in the decision, but that they’re exactly balanced on either side. At this magical restaurant where I’m doing my example deciding, I could get tacos which I just had yesterday but they still sound good (but won’t be as good since I just had them), or I could get Chinese, which will I haven’t had in ages and also sounds good, but promised to have again with a friend (and if I had it now, might ruin my appetite for it later). This way there are reasonably similar arguments for and against each, and I will weigh them carefully before deciding. This is comparable to the original emotionless argument in the sense that it’s still totally balanced, but this time once a decision has been made, I’ll have to live with the emotional repercussions of that decision (the friend being annoyed with me, perhaps). Therefore, when I’m making the decision I’m taking the emotions into account. Why couldn’t a machine do the same thing? If given the ability to process the ideas of emotions similar to that situation—ideas being the key word here because I sense that my dissenters would say at this stage that a machine couldn’t process “emotions,” but I think it makes sense to say that could handle “ideas,” just as Deep Blue knew the idea of chess and could rationally make a move—it would have the same difficulty (not necessarily with eating, since last I checked even AI machines don’t need to eat). And then, to make an exciting point, if they could weigh the decision as carefully as I could and take the same emotions into account, that would mean they would have to understand emotions. And to understand emotions is to feel them, since no amount of description will allow someone to feel a given emotion. And ta da, in a proof-y step-by-step, mostly logical way, AI could really exist in the thinking-feeling-caring kind of way.