General AI Challenge (2017)

What is wrong with the Hutter Prize strategy for achieving AGI? Don't get me wrong - the money would be very nice to have - but I am drawn to the simplicity of the Hutter Prize. Is the conception of that prize in some way flawed?

I think that finding an efficient way to invert mathematical functions gets you most of the way to AGI. What we need is a general purpose declarative constraint satisfaction programming language. Take the task of finding the square-root of a number in a system that initially knows how to square a number.

In my Lisp like language:

(define).. (square). x (*).. x x

(write). "sqrt(9) = " (write). (square). () 9

;finds the number which squared equals 9

;This program will figure out the answer by inverting the "square" function. In general you can specify any constraints that you wish and then solve for the unknowns using genetic algorithms or some other general purpose optimisation technique