My man, you’re speaking sci fi, not what we currently have.

The biggest of current LLM models contains ~ the same number of parameters as we have neurons. It’s not a 1:1 mapping because parameters are closer to neuronal connections, but from a pure numbers standpoint we are operating at the scale where we can start creating true simulated intelligences, even if not human scale just yet.

This doesn’t mean current LLMs are that intelligent, just that it’s not sci-fi to think we could create a simulated intelligence now.

Furthermore, both philosophically and materially, the notion that consciousness cannot be computed is more than gaining traction.

Is it? Do you have any sources / do they have any explanation for why neurons can’t be simulated?

If humans ever make something with free will and volition, something that isn’t just doing things on command but has its own wants, sure. But we might never get there, and that’s a real possibility. Intelligence isn’t in solving equations but in imagining the math problems.

I mean, we’re talking about whether or not an AI could make music. If it creates a new song, with lyrics and music / a melody that never existed before, and people listen to it and sing it and dance to it and enjoy it, how would it not be music?

AI is not “making anything”, it’s regurgitating combinations of previous stuff on-command.

Even current day LLMs are doing more than just regurgitation, even if they fall far short of human intelligence.

And at a fundamental level, there’s no reason to think that simulated neurons running on computer chips can’t be as intelligent as us, if we can figure out the right way of wiring them so to speak.

There’s no inherent law of the universe that says that only biological humans can be intelligent and can thus create music.

That’s not a redefinition, lol, music is a human construct. Nature has lovely noises and birds chirp, and by itself, even if it constitutes notes and waves, it isn’t music.

A gorilla or ape can’t sing or make music? Could a neanderthal? Homo florientis? Homo erectus? What is it specifically about homo sapiens that give us the unique ability to make music and sing, that no other animal has?

Again, if you predefine music as being made by humans then you’re not engaging in a discussion or logical debate, you’re just arbitrarily setting goal posts to guarantee that you’re right.

People need to get over the idea that algorithms can’t be intelligent because they’re algorithms. Algorithms can model the behaviour of the neurons in your brain, meaning that they can model your brain and intelligence. We are obviously not there yet with LLMs, but just saying ‘numbers and math = not intelligent’ is quite frankly dumb and just shows that you don’t understand math, physics, biology, neuroscience, etc.

There is no human intentionality in AI-created music.

The entirety of your argument boils down to you arbitrarily deciding that music needs to derive from human intentionality.

That’s not an actual argument about whether or not AI is capable of creating music, that’s you redefining music to make sure the answer is no.

“Drawing the line” is a Thought Terminating Phrase, which is a concept worth reading and thinking about.

In the abstract / in your example, if you’re just “drawing the line” then no, that’s not a legitimate argument. It is literally just you saying “nope, I arbitrarily say this is different then this”.

However, if you can back up why one thing is different from the other, then it is valid to distinguish between them. Sometimes it can be worth dividing a system into chunks and drawing arbitrary lines rather not drawing any, but you should still be able to logically back up why it’s better to chunk things than not.

But in your example, it sounds like you don’t actually have a logical argument, just one based on you arbitrarily deciding that music can only be made by a human.