Cogito ergo sum.

Accepting a common framework of provable, i.e., measurable, repeatable, falsifiable phenomena, as a concept of “reality,” seems to be a pragmatic approach, given my sensory inputs and the processing results of my brain. This is then “knowledge.”

But ultimately, this is subordinated to the possibility of an illusion – be it like in The Matrix, or as a Boltzmann brain, or whatever. Unless there is evidence for that, it appears most practical to me to go with the above, as I don’t gain anything from racking my brain about such possible illusions of reality (even though it’s fun thinking about it).

I’m nitpicky about the word “believe”. So let me rephrase: I do not believe. Either I know, or I don’t know. Everything else are more or less informed speculations, assumptions or hypotheses at best.

Ask for a community meeting, so you can see that those people are real.

Despite that, I don’t see any effective counter measure in the long run.

Currently, sure, with a keen eye you might be able to spot characteristics of one or the other LLM. But that’d be a lucky find.

Yes to all of that except your last paragraph.

I suppose you’re referring to the article I’ve linked. As I see it: If an increasing amount of applications world are running with Python, then energy and time consumption are important aspects. Not only cost wise but especially since we’re grilling our planet. Therefore, comparing with more efficient languages is indeed meaningful.

Python sucks.

Not only is it extremely inefficient, it is also a pain in the ass to work with if you have to use APIs that heavily rely on dynamic type wrapping and don’t provide stubs. Static analysis via Pylance is not possible then and you’re basically poking around in the dark, increasing the difficulty enourmously to get to know such an API. Even worse if there isn’t even a halfway decent documentation.