defending the unthinkable
i went to college with the intent to major in computer science and design a computer that thinks. pretty ambitious, but computer power was and is flying up and after all, kurzweil says it’ll happen in 2030 or whatever is his hypothesis now, so it would be in my lifetime. as perfect time as there could be for it.
along the ways i ran into a single quote from the 70s that completely stopped me.
“The question of whether a computer can think is no more interesting than the question of whether a submarine can swim” – Edsger Dijkstra
in one sentence it summarized for me that “thinking computer” is a just a language illusion. it’s not real and it will never be real.
a computer will never think in the same way a submarine will never swim. swimming is what living things do, a submarine instead ‘propels itself through water’. a computer will never ‘think’, it will ‘perform computations in order to arrive at conclusions’. which, guess what, they already do and have been for ages.
for people to agree that a machine is a “thinking computer” would be one that can feature a display of a pleasant cartoonish face that, when computing, would furrow it’s brow and make “hmm” noises. while an interesting task and a cognitive/behavioral challenge, it’s not a computer science problem.
the better goal i learned in compilers: we should be working to precisely define problem spaces where computers can help with decision making, and then writing better and more robust expert systems (by whatever buzzword they’re going by nowadays) that can read data about the situation, and suggest or perform actions in response. not as glamorous as ‘thinking’, but infinitely more useful.
