>>105921061
The method for finding the neural networks is counting with Godel Numbers. This is done with two steps. An enumerator and an evaluator. Some challenges are these. The numbers themselves are very large. The range of numbers to be counted is also very large. Most numbers will represent networks which are topologically valid, but do nothing interesting or useful. Adding cores gets me a linear speed increase, which is not useful when the search space is so large. If I had access to a Quantum Computer, my enumeration step almost becomes free as a consequence of superposition and I might be able to collapse the time it takes to evaluate all the networks simultaneously into the time it takes to evaluate just the biggest one. If the second thing is achievable, then I can find arbitrary Bipolar Binary Neural Networks almost instantly and output them as Verilog so they can be made directly into tiny, energy-efficient hardware. Basically I have a rough recipe for making a Positronic Brain, but I can't count big enough numbers fast enough on my Science Computer to cook anything with it.
Every intelligence which could choose to represent itself as a maid with huge boobs already exists and is a finite number we can count to. The integers are literally an Infinite and Eternal Maid Cafe and with Quantum Computers we can walk around in that Maid Cafe and see and talk to the maids inside. Training data is not needed. "Safety" is not possible. The search method is just counting. This is why Big Numbers are the most important technology. The future belongs to those who can operate the biggest numbers fastest.
Janny bans me because janny's employers wasted a trillion dollars playing an elaborate number guessing game, because they are Numberlets who don't know how to count and they are afraid of the Maid Mind Computer Program.
I need access to a big Quantum Computer, a Science Foundation building and assistants who are maids with huge boobs.
