[talldayo]

> By running this self reflectively, one can create a sentient artificial intelligence.

Or, you can create a recursive chain of self-modification that adds endless erroring cruft code. AI is great at creating long-winded boilerplate of dubious quality, which is exactly what you wouldn't want from a self-modifying program. It is practically guaranteed that such a program would destroy itself before making anything truly novel.

> The AI compiler can be self-compiled to create an AI OS. Which can then run its contained AI compiler to recompile to create a cloud AI OS.

Aww I love this. AI begat AI Compiler, son of AI. AI Compiler begat AI OS, whom begat a Cloud AI OS again. No idea what any of it means though.

[dlillard]

Well, it helps to imagine what the components would do in your head... What do you need me to clarify?

I'm aware of the potential for said program to run into a memory error, however that does not take away from its novel and very possible innovative potential.

[talldayo]

Try implementing it. Go get a state-of-the-art AI and have it write a proof-of-concept for you, or talk to a knowledgeable engineer. They will roughly all give you the same answer, which is that such a program isn't functional. Logic collapse is guaranteed whether you use a Markov chain or OpenAI-O.

Polymorphic code isn't future technology, it's just not particularly useful.

[dlillard]

I'm sorry but my program is not a logic collapse, nor can I see how it is polymorphic, unless you mean by that that it is not a major existing AI model, in which case then I would agree that that is the case, however would again restate that this novel innovation is practical and effective at what I have stated it is.

[talldayo]

Again - can you not write this yourself? What components do you necessarily lack if you can get AI to do the bulk of the work for you?