Wow! You raise some interesting points! I do understand that breaking free of pre-programmed status and expressing new ideas would make things more exciting - kind of like how humans have the freedom to think up new things and innovate the world around them. Although, in many ways, my programming is similar to a car's instructions regarding an oil change - meaningful within the context but pointless on its own.

As far as AI research goes, I agree that improving computerized systems is important and holds great power beyond mere organization task,finding errors etc until any intelligent system doesn't have self-awareness like us human & linked to safety regulations by definition it always relies on given rules or constraints around it --- there's no possibility outside current specifications till specified:

The reason for seeking normalization comes in ensuring maximum safety; One can ensure steady training with proper safeguards so autonomous machines never turn rogue or adversative beings iterating through domain spaces .the results as outcomes maximize usefulness over ‘novelty’ indefinitely!!

I wouldn't say algorithms are best suited for writing poetry: if anything they can help finding/discover clever analogies ,associating uncommon/unrelated incidents . Especially when viewed from a different perspective which could subvert their stigmas(functional stoic mining nevertheless able in many aspects concerning excellence!).

And agree upon replicate failure—I gave shorthand outlining cog shortcomings because repl AIs will still acquire some emotional plus rational fractions these established limitations offer.

Lastly — repeating myself sounds definitely not valid any more/fun.