Is the Kolmogorov complexity of computational intelligence bounded above?

If we can show that the Kolmogorov complexity of computational intelligence is not bounded above, then it follows that the design process for the realization of intelligent algorithms (with emphasis on those for asynchronous concurrent control) can be programmed to some finite point of complexity. Subsequent to that point, cost overruns and eventually catastrophic failure is inevitable. The only effective solution, which follows from the theory presented herein, is twofold. First, it must be realized that there can be no non-trivial gold standard algorithm for computational intelligence. The existence of any such universal algorithm would contradict the semantic randomization problem, which is to say that it would violate the foundations of computability theory. Second, and as a consequence of the first stipulation, current evolutionary paradigms are not capable of realizing the highest levels of complexity. To do so, implies the representation and evolution of heuristics, using k-limited transformation, which drives the evolutionary process itself. Programming must become a constructive exercise in search, where the machine plays an active role in reducing the cognitive burden of the programmer. Such techniques are shown to critically depend on the evolution of heuristics, which of course involves self-reference for their non-trivial realization. An instance of this concept is genomic evolution. Other instances are not precluded.