Do computers think?
I don't know. Do thermostats think about the temperature they regulate in a home? Perhaps they do. I do not think that panpsychism is a ridiculous philosophical position at all. Perhaps rocks think thoughts like: "It is so nice just to lie here. I really hope I can just lie here for… Oh crap! Some kid just picked me up and is throwing me."
The point I have been trying to make is that the fact that a machine does the things we built it to do is no evidence that it is thinking. And the fact that we can build a machine the behavior of which sometimes surprises us does not alter the fundamental situation in any way: One could build a thermostat that changes the temperature based on some complex relationship of its setting, the present temperature, air pressure, noise in the house, and humidity level, and we often might be surprised that shouting causes our house to become very cold. It can be very difficult to guess in advance how some combination of these factors might interact to raise or lower the temperature. I recently programmed a predator-prey model with over a dozen parameters I can set: I am often very surprised at how tiny changes in one parameter can produce wildly different behavior in the system. But I think this is simply because I do not fully understand the machine I have built, and not because the system has got into its head some new idea on how to behave.
Does a toilet "know" that when we push the handle, we want our excrement to go away? Obviously a computer is a much more complicated machine then a toilet: but how is the fact that we built a very complex machine suddenly evidence that the machine is thinking?
UPDATE: And note that thermostats do something it used to take thought to do: in earlier times, when it was too cold, someone had to think, "Hmm, guess I'd better go put more wood on the fire."
The point I have been trying to make is that the fact that a machine does the things we built it to do is no evidence that it is thinking. And the fact that we can build a machine the behavior of which sometimes surprises us does not alter the fundamental situation in any way: One could build a thermostat that changes the temperature based on some complex relationship of its setting, the present temperature, air pressure, noise in the house, and humidity level, and we often might be surprised that shouting causes our house to become very cold. It can be very difficult to guess in advance how some combination of these factors might interact to raise or lower the temperature. I recently programmed a predator-prey model with over a dozen parameters I can set: I am often very surprised at how tiny changes in one parameter can produce wildly different behavior in the system. But I think this is simply because I do not fully understand the machine I have built, and not because the system has got into its head some new idea on how to behave.
Does a toilet "know" that when we push the handle, we want our excrement to go away? Obviously a computer is a much more complicated machine then a toilet: but how is the fact that we built a very complex machine suddenly evidence that the machine is thinking?
UPDATE: And note that thermostats do something it used to take thought to do: in earlier times, when it was too cold, someone had to think, "Hmm, guess I'd better go put more wood on the fire."
What do we know thinks? People. As far as we can tell they think with their brains. Better: their brains think. As far as we can tell they think by processing information. Better: they host electro-chemical operations on electro-chemical states. Stop me when you see where this is going.
ReplyDeleteMaybe all our computers are determinate and always will be. But then, maybe so are we. The determinate machine doing your determinate calculation is still sensibly describable in terms of its mental contents, not its atoms. Thst it seems to me is the real meaning of thinking: comprehensible (only) in terms of information and representational states like beliefs, rather than by differential equations of motion.
Hey, you know what else "hosts electro-chemical operations"? Thunderstorms! So I guess thunderstorms think! And my toaster too! It knows when my toast is just right! And I can sensibly describe my thermostat in terms of "its" mental contents, so my thermostat thinks too.
DeleteOr maybe "its" is a terrible mistake here: maybe it is always *our* mental contents, that we project out onto objects in the world.
And let me again insist: I am NOT saying computers don't think. But I AM saying every reason we have to assign thought to computers would also lead us to assign it to thermostats and hydraulic computers and toasters.
DeleteAnd note that thermostats do something it used to take thought to do: in earlier times, when it was too cold, someone had to think, "Hmm, guess I'd better go put more wood on the fire."
ReplyDeleteyeah... but adam smith wrote of the kid, hired to move a bar that moved a wheel, and the kid tied a string from the bar to some other gizmo and let the string do his work for him.
it seems the thought is captured in the technological advance.
beyond that, i got nothin
Yes, these things embody our thoughts.
DeleteI get what you're saying Gene, but don't ever get in a spaceship controlled by the HAL-9000.
ReplyDelete