And how is this connected to the thing I was responding to, which was "Our cosmos is shown to be discretized (quantized chunks)" (which is a retrodiction)?
Maybe I should rephrase the question. Describe in some detail a realistic experiment that we could run today or in the very near future that could be used to test your model on a manageable scale. "Create a human level AI or a universe" is neither realistic in any sensible time frame, nor remotely detailed enough to take seriously.
The redefinition sequence yields classifications of the redefined god, such that:
(1) humans are empirically minimally capable gods, as they can engineer human level intelligence (not ai, but humans can engineer smarter instances of themselves by learning tasks)
(2) humans may become highly capable gods by either engineering non trivial ai, or yield universes. (This is the part that is a prediction, through modern sciences such as digital physics or Penrose Hawkins' singularity theorems)
While prediction (1.b) is well, a prediction, we see evidence that universe yielding or non trivial ai yielding is probable, rather then not:
(a) Machines now run efficiently (though still not human- efficiently) at 10^14 + simulated synaptic ops per second, of the total human 10^16 to 10^18 synaptic ops per second. Notably, as efficient parallelism increased over time, machines did more and more cognitive tasks. (and machines continue to enhance with better artificial neural synaptic approximations)
(b) We are simulating more and more detailed universes. (Eg illustris) As time had passed, simulations had enhanced. Science deals with predictions, and there is no reason to ignore the progression of capability/time relation.