This UD post got me to thinking. I do that from time to time.
On the subject of computational intelligence I have some minor credentials, including a Silver Medal at the first Computer Olympiad in London, sponsored by David Levy of chess fame. You can access the final results of my research and efforts in computational artificial intelligence (AI) here:
If you have a computer with sufficient memory and disk space you can explore the only perfect-play endgame algorithm ever invented for the game of checkers (known as draughts in the UK).
It was my exploration into computational AI that initially caused me to have doubts about the creative powers of the Darwinian mechanism, which I now consider to be a transparent absurdity as an explanation for almost anything of any significance, and certainly not as an explanation for human intelligence.
Here’s why.
Computer programs that play games like checkers and chess involve two primary algorithms, a brute-force tree search and a leaf-node static evaluator. The tree search says, “If I move here, and the opponent moves there, and I move thus…” Unfortunately, the exponential explosion of possibilities means that the search must eventually be terminated. At that point a static positional evaluator must be invoked. This requires designing heuristics that can evaluate the position with no further search.
The problem is that these heuristics are difficult to devise and encode, and they are often wrong, because of tactical considerations that lurk beyond the horizon of the search and the fact that the heuristics can have unanticipated side-effects.
A human player might say, “Hmmm, if I move here, this will create a positional weakness from which the opponent cannot possibly recover.” There is no way to encode such knowledge, which comes from human experience and positional recognition.
The other problem is that computer programs like mine do not play against the opponent; they play against themselves. The tree search assumes that the opponent sees everything it does, which in a human-versus-human game is not the case. In one game my program played against a grandmaster human, the human was in deep trouble, and he told me so. The computer was considering the move the human feared, which would lead to a very difficult, razor-thin draw. But the program searched so far ahead that it found the draw, assumed the human would see it as well, and played a move that gave the program a few more meaningless points, letting the human off the hook.
All attempts at computational language interpretation have been dismal failures for similar reasons. Even the best spell- and grammar-checkers are astronomically stupid:
Eye halve a spelling chequer
It came with my pea sea
It plainly marques four my revue
Miss steaks eye kin knot sea.
Eye strike a key and type a word
And weight four it two say
Weather eye am wrong oar write
It shows me strait a weigh.
As soon as a mist ache is maid
It nose bee fore two long
And eye can put the error rite
Its rare lea ever wrong.
Eye have run this poem threw it
I am shore your pleased two no
Its letter perfect awl the weigh
My chequer tolled me sew.
The point is: With all our human intelligence, technology, and inventiveness, how can anyone who is still in contact with reality believe that random accidents engineered our brains and minds?