The parallels between your critique of music analysis, and linguists' critique of LLMs, bear remarkable similarities. "Language/thought is more than sequences of tokens" will still be true no matter how much data we throw at the problem to smooth the rough edges.

▲

Levitz 8 months ago | parent | next [-]

The parallelism doesn't really work, I'm going to try to stretch it to make a point though.

Imagine that we were at a stage in which LLMs didn't really make sentences, only output like "Potato rainbow screen sunny throat", then we studied which words are used. There's really not much value to the words at all, we could maybe see which words are bundled together, we could try to ascertain what kind of words are used more, but in wanting to study the coherence of it all, it just holds very, very little value.

Chords by themselves hold very little meaning. The sensations evoked come from chords in a context and the progression provides very valuable context. Talking about a chord in a song is like talking about a word in a book, it's never really about that piece of the puzzle appearing, it's about how that piece is used in the puzzle.

	▲	uoaei 8 months ago \| parent [-]
		It does work, particularly the emphasis on causal sequences being wholly inefficient to represent multidimensional and abstract concepts such as those that exist in both language and music. The fact that you never refer to "syntax" even in this attempt at high level reasoning gives me pause and I cannot help but to conclude that you are making arguments in bad faith.

▲

airstrike 8 months ago | parent | prev [-]

Except music theory has a math component to it so it's arguably somewhat quantifiable and falsifiable in a way that linguistics never will be.

▲

Tainnor 8 months ago | parent | next [-]

There is a lot of mathematics that can be used to analyse language. Phonetics is basically acoustics, phonology has things like optimality theory, for morphology you can use finite state machines, syntax uses formal grammars, statistics obviously plays a big role in certain areas, etc.

Both in the case of music and linguistics, there are people who argue (probably not wholly without merit) that looking at the mathematics too much is missing the point.

▲

uoaei 8 months ago | parent [-]

A refusal to acknowledge such integral parts of these systems as semantics renders that line of argument wholly irrelevant. No good faith discussion cannot be conducted without participants who already understand the meaning behind symbols.

▲

Tainnor 8 months ago | parent [-]

I'm afraid I don't understand your reply.

	▲	uoaei 8 months ago \| parent [-]
		It doesn't matter at all what "mathematics" you use to "analyze" music because the purpose of these modes of communication goes beyond the arbitrary sounds chosen -- what does matter is what is being represented by those sounds. Statistics is for the analysis of systems in equilibrium. Language by definition is a non-equilibrium system.

▲

uoaei 8 months ago | parent | prev [-]

[flagged]

▲

airstrike 8 months ago | parent [-]

This is an incredibly ignorant reply, even for someone who has so obviously studied neither debate nor logical reasoning.

I encourage you to read this site's guidelines and strive to do better moving forward, as your comment kinda seems to break 5-6 rules all at once.

https://news.ycombinator.com/newsguidelines.html

▲

uoaei 8 months ago | parent [-]

If you had more to say on the topic at hand, you would have said it by now. I'm not interested in trying to educate someone so comfortable in promulgating assertions informed by nothing but amateur-level vibes.

We could list the logical fallacies you're displaying in the spirit of "debate and reasoning" but I have no faith this would go anywhere productive.

Report me instead of complaining about guidelines if retaliation for being exposed for your ignorance is so important to you.

▲

airstrike 8 months ago | parent [-]

But it would be terribly wasteful to further entertain someone on the subject of music and artificial intelligence, when it is so evident they must first overcome their oblivious impertinence and emotional incompetence in order for such debate to be even moderately enjoyable to anyone else less arrogant and unhappy.

You assume my choice to not continue the conversation is because I have nothing to say, when in fact I do so because I have nothing to say to you specifically.

	▲	uoaei 8 months ago \| parent [-]
		Oh good, you know big words. Next on the list is to learn the relevant big words for the things for which you claim competence. I'll wait. PS: it's funny that your best comeback is just to imitate the critique I levied against you. You could demonstrate your "debate and reasoning" ability by crafting an actual argument, but you don't. I'm convinced that was not a cognizant choice but an act of desperation to save what's left of your ego. I've dealt with your type enough to know what you are, and after all this you will be forced to acknowledge the same... unless your ego really is strong enough to overcome your capacity for "reasoning".