I’ve already outlined the model. This really was just a test of Lesswrong. I think what I’ve said will be an obvious accepted truth in the future. Either once we’ll get AIs capable of deducing contrarian things from first principles or once our society grows up a bit.
I don’t want to argue my side more tbh but what you’ve outlined is just one example of a very rare usecase of music which really doesn’t disprove anything.
vire
You don’t need empirical evidence to reason from first principles. In math we do that all the time.
I’d expect that people listening more to music might even be better verbally. But that’s simply because their brain is more oriented for this language learning which also causes more rewards from music. So correlation not causation.
This is basically my outline for proof:
something gives humans value → it gives us pleasure (evolution in most cases tried its best to make sure this is true)
it gives pleasure → something gives human value ∨ exploits some property of rewards system that generally in ancestral environment resulted in value
music gives pleasure, now i will show it exploits some property of reward system and we all known the disproportionate evidence to its value given how much pleasure it gives, so since we know it doesn’t give value it is already reasonable to suspect that it exploits our brain, but i will further drive this point by showing how exactly does it do that
our reward system especially in humans evolved rewards for predicting audio tokens, eg. to learn speech more effectively, music is mostly about lots of very easily predictable impactful audio tokens, this model of music already explains why we like similar genres, why music usually sounds better after second listen, explains its structure, probably many more things
we have seen in AIs just how effective learning mechanism prediction of tokens is
Musicians would have to near incoherent to talk to
Not that severe...
However speaking as someone with success on informatics olympiads (requiring verbal reasoning) I do believe that avoiding music can give some useful edge.
In context of humans it’d probably make sense to define value as improving our inclusive fitness (the evolved utility function). But it could also be many other things, for instance money made in one’s lifetime.
Yes if it’s that or something drastic that would indeed be value. But would you say that smoking weed therefore gives value in general? I wouldn’t. Also oftentimes our brain learns to reach these emotional states because we reinforce it with the follow up reward.
If you’re looking for studies then the thing with contrarian claims is that they exist because so far there weren’t any.
However I can give out a logical argument for why that’s the case.
Humans use language to communicate, reason, learn, etc. Music exploits the reward circuitry responsible for learning language (I can further give proof for this). By listening to music you are severely limiting your ability to do or improve in anything language related. Because the reward circuitry is “exhausted” from prior listening to music.
And yes there are “side effects”. Just harder to observe.
I said the similarity is in “all pleasure no value”. Pleasure is an evolved signal for value usually, but oftentimes it can be exploited eg. with sweets too fatty foods. So saying there is “value in pleasure” doesn’t make a lot of sense.
Following pleasure signals like the activities you have shown (besides cats) is usually a good idea. Altho sometimes the pleasure can be simply explained as with drugs as some form of exploit of our imperfect brains, in which case there is indeed no value.
In fact I challenge anyone here to disprove my claim. If you want I can lay out my proof for the claim as well (altho I feel this community is a bit undeserving given the immediate disagreement)
Guys this was a test of Lesswrong on how it will respond to obviously true contrarian statement. Thanks for proving my point that people on Lesswrong aren’t actually that rational.
Music is just like meth, cocaine or weed. All pleasure no value. Don’t listen to music.
That’s an interesting point, why didn’t we see major improvements in LLMs for instance when coding… Despite them achieving reasoning on the level that allows them become a GM on codeforces.
I’d say this is a fundamental limitation of reinforcement learning. Using purely reinforcement learning is stupid. Look at humans, we do much more than that. We make observations about our failures and update, we develop our own heuristics for what it means to be good at something and then try to figure out how to make ourselves better by reasoning about it watching other people etc…
This form of learning that happens at inference time is imo the fundamental thing preventing LLMs right now becoming more intelligent. And actual memory of course.
So we’re just making them improve at measurable tasks through naive reinforcement learning but don’t allow them to generalize it by using their understanding of that to properly update themselves in other not so measurable fields...
I’ll understand if this offends some people here who are researchers and don’t have much profit but I’m assuming a functioning society where good research is adequately rewarded.
How to fix universities: make their profits tied to competency of leaving students by taking a percentage of their future profits for the next x years.
The new Dwarkesh AMA episode is really great. Would recommend everyone to watch it.
vire’s Shortform
Does listening to music regularly lead to lower probability of achieving goals?
Alright @dirk I’m happy to make a bet since you think it’s less than one percent I’m going to be generous and give you 8:2 odds. Eg. you put up $800 I’ll put up $200. And once we’ll have truth seeking AI whether it will confirm that my model is correct.