Conversation with Assistant about how to have computers put unused resources to improve own code, plus howto program Arduino/Elegoo.

Is about a neuromorphic default mode network, plus robot reverse kinematics/computer vision.

This post allows all uses.

Conversation with Assistant is: https://poe.com/s/Sieywy601Jb38SZikLxh

The first and last messages are about robot inverse kinematics/computer vision, the rest is about howto produce a computer (with minimal bootstrap code) which uses unused resources to produce/improve its own code. The minimal bootstrap (which humans produce) is just what the computer must have to access its own compute resources (such as simple disk access, memory use, use of numerous CPU cores, plus an abstraction somewhat above assembler (such as C++ or LLVM Intermediate Representation) which allows the computer to produce code (which it builds into programs) which will execute on most computers).

Above this, humans would give it some sort of fitness function(s), such as how much humans interact with the computer (or how humans score what the computer does.)

Thus the computer would use unused resources to plan both; what sort of programs to produce, plus how to improve those programs (to reduce resource use, or to improve what those programs can do).

The conversation with Assistant includes examples of how the human Default Mode Network allows humans (before entrance into formal schools) to use unused neural resources to improve how humans compute, with a request for Assistant to assist to produce computers which do so. Assistant responds with somne pseudocode but of course can not produce all of this.

Was about to update first comment but won’t since you already answered. *Upvotes*

But perhaps more practical than to mimic human neural tissue (as Default Mode Network), is to produce an iterative (not perfect on first run, but can improve through short iterations of a loop which uses background resources) version of Superoptimization - Wikipedia which can produce code with loops (Wikipedia’s article just describes an algorithm for loop-free code, but lots of formulas must use loops).

Coupled with a speculative (rather than require descriptions of the desired programs, do analysis of usage to guess which new programs to produce) version of Program synthesis - Wikipedia.

Have so much to do now, but if noone will produce this, will have to produce a combo of those 2.