Written: Sept. 22, 1996
I just finished reading the extremely interesting, well-written, thought-provoking, and provocative book ``After Thought: the computer challenge to human intelligence'' by James Bailey (Basic Books, 1996).
The main point of the book is that the human way of thinking `sequentially' is (going to be soon) dead, long live `parallel thinking', based on empirical computer simulations where the computers will teach themselves, evolve naturally, and soon surpass us feeble humanoids.
While it is true that thousands of years of managing with this tiny computer between our shoulders gave us some bad habits and hangups that we should learn to get rid of, it would be stupid to abandon the Cartesian tradition altogether. Besides, labelling the human way of thinking as sequential is a gross over-simplification. All the great breakthroughs of the past were achieved by making connections, and thinking in parallel. Only the verbal presentation of the end results (i.e. proofs) seemed sequential.
The beauty of Genetic algorithms should be carried to the meta-level. The interbreeding of computerish and humanish thought-processes would be a real blast, and would lead us to peaks unreachable by either alone.
In the future one would be able to find genetically not just a sorting algorithm to sort 8 numbers, but n numbers, and even a proof of the Riemann hypothesis, since after all a proof is just a collection of bits. But, in order to do this, we should preserve all the human proofs of the past and teach the computer how to emulate them, and ultimately surpass them.
The successful mating of the Genetic and Cartesian approaches to thinking would be most efficiently achieved with computer algebra, which is high-level, object-oriented, rather than low-level. So symbol-crunching rather than bit-crunching would lead the way. So both Optimists and Pessimists, learn neither Russian, nor Chinese, nor Fortran, nor C, but Maple.
Doron Zeilberger's Homepage