Rick Rubin on AI (& now Graeme Revell, too)

[Zeisner]

Again, I think we give our own level of free will far too much credit.

I suggest to leave free will for philosophers to talk about, that's their thing. It also has nothing to do with the entire AI topic (like pretty much almost everything that is mentioned in any AI thread).

IMO we're not that far off from stochastic parrots.

How can you know if there isn't a complete model yet? And considering that not all mechanisms known from natural neural networks have been replicated in hardware and software yet? Which is a fraction once again? What we got so far in AI are proto neural networks. Considering the evidence available it looks like the difference is larger than previously expected (based on the neuroscience boom that started in the 90s).

[jancivil]

This hit my feed I think on Saturday. They're divvying a longer panel discussion into shorter videos on Youtube. This directly followed the clip I'd seen previously. The whole of it can be seen in a free trial of the IAI site.

As pertains to this thread, the title they chose is ironic. The title doesn't really reflect the pertinent focuses in this, a little bit click bait. No one is giving any time to "Free Will", the only point of 'psychopath' is that this too is an example of consciousness...

It's time to pay attention to people that know what they're talking about.

[jancivil]

Sentience means will. "Free Will" is a big red herring.

Mostly I get being skeptical, AI is not sentient, yet, but the idea of sentience is already loaded with false narratives and vain conclusions. IMO it will not take much more to make something like sentience a rational conclusion to draw from AI computer patterns.

Loaded with false narratives like what? I've not presented any. What does that gesture have to do with the discussion? Other than you seem to be trying to support a viewpoint and are grasping.

"something like sentience" - that isn't doing the work I suspect you want it to. Sentience is not an ambiguous or vague concept. NB: the LLM is already fabricating simulacra. "Life-like", "something like sentience"... Sentient means aware of the experience of consciousness, can feel and sense, IE: has a definite experience of qualia. Frank Jackson's qualia argument aka Mary's Room is essential reading. Mary has access to every known fact about the color red but has never seen it. For her to know what it is, she has to know what it's like; she has to leave her room. The experience of the color is not theoretical.

How you gonna make LLM know what anything is like? It is not conscious. Penrose: consciousness is not computational; the conscious mind isn't sussed by physicalism, it's no more a brain state than waves of an ocean are discrete components of the ocean. How do you get around this?

No one knows what causes consciousness, so how are we suppose to impart it to something that doesn't have experience to be aware of? You're asserting it can be coded. I have no reason to believe that.

[machinesworking]

Again, I think we give our own level of free will far too much credit.

I suggest to leave free will for philosophers to talk about, that's their thing. It also has nothing to do with the entire AI topic (like pretty much almost everything that is mentioned in any AI thread).

IMO we're not that far off from stochastic parrots.

How can you know if there isn't a complete model yet? And considering that not all mechanisms known from natural neural networks have been replicated in hardware and software yet? Which is a fraction once again? What we got so far in AI are proto neural networks. Considering the evidence available it looks like the difference is larger than previously expected (based on the neuroscience boom that started in the 90s).

Everything in life that I've seen so far, from neurobiologists, psychology etc. leads me to the conclusion that the human brain is amazing, life is amazing, but it's not special. The thing that you get amazed at, complex neural networks etc. are not beyond replication. What is will if it's not free? it's just a pattern then, X person has a pattern of being able to break a cycle, or stress a physical limit. Grit, determination, empathy etc. all of those things are behavioral patterns based on DNA and nurture, the environment your particular brain was lucky or unlucky enough to grow into adulthood in.
So if you strip the mystery away, and look at it as a complex pattern, then replicating it is not impossible. I do realize though that this will seem "reductivistic" but I don't really see the need to go beyond the simple fact that mankind overvalues it's own complexity, to the point to where we're in the process of developing helper programs that emulate human interaction and already those programs are transgressing conventional morality because the developers choose to allow CSAM, or trolling behavior with "Mecha Hitler". It really doesn't matter whether these things have sentience or not, if they randomly emulate transgressive 4chan troll behavior and at some point start being asked to make decisions based on those, then the result will not be determined by the programmer anymore.

I mean again, decisions are made before the conscious mind is aware of it, this eliminates free will, and will isn't anything but a predetermined product of environment and DNA, then what determines sentience?

[Zeisner]

The thing that you get amazed at, complex neural networks etc. are not beyond replication.

Nobody can replicate what is not known (yet).

So if you strip the mystery away, and look at it as a complex pattern, then replicating it is not impossible.

Depends on available technology and ultimately physics. Look at the possible permutations of sets of neural connections for example. You would need insanely powerful computers to replicate that. More powerful than any science fiction writer has ever dreamed about. Calculating the current state of every single atom in the entire universe would be easy in comparison. So even if we had a complete model, it's very unlikely that we can do much with it, at least when it comes to AI.

I mean again, decisions are made before the conscious mind is aware of it

This does not apply to higher thinking though. Math is a good example, you can't just follow some instinct or feeling hidden behind the mask of consciousness to decide if an algorithm works or not.

[jancivil]

Everything in life that I've seen so far, from neurobiologists, psychology etc. leads me to the conclusion that the human brain is amazing, life is amazing, but it's not special.

So you're dismissing that consciousness is special? Actually no, in fact you only retreated to talk of the brain. All you've done is gainsay the central point with no consideration, providing only assertions with zero support.
"So if you strip the mystery away, and look at it as a complex pattern, then replicating it is not impossible."
Does avoidance of the mystery strip the mystery away? Who has done this, stripped the mystery of consciousness away, this central and crucial problem creating a form of intelligence presents? There is a definite answer here, no one.

In sum: I cite Roger Penrose who's stated that the brain is computational, along with "consciousness is not computational"; in support of this we can note that to all our understanding consciousness has no apparent location - as a point of fact, no one can demonstrate any different - example given here is_the_brain. What comes back is "the brain is amazing but it isn't special". There's only loss of my time for me to endeavor to communicate and be utterly ignored like this.

I've said all I can, you have a belief that's keeping you from understanding. The belief itself is absurd. To hold onto a belief with no evidence of any kind in support of it - at the least you definitely have yet to begin to provide any - borders on a delusion.
No mas.

[jancivil]

Biological life emerged from vast numbers of independent, self-sustaining physical interactions with intrinsic reproduction and selection. AI systems, even at massive scale, which this is not, sample from fixed, externally defined distributions. Increasing participation raises output volume, not autonomy, persistence, or evolutionary dynamics. None of the bots are independent in any reasonable sense in this analogy. This is a puppet, not a boy.

Even if you reduce the claim to that of autonomy, this is not that. It seems that the casual hype has more than a few of you fooled into thinking that this is really something.

[...] in your own argument there are serious flaws. Biological life does seem to indeed be the product of random natural occurrences, that we can agree on, ...

He states "... interactions with intrinsic reproduction and selection" - argument from Darwinian evolution - but now we can agree on what, again?

Darwin used the term "chance" to describe variations in organisms, which he later clarified meant a lack of understanding of the causes of variation, rather than true random, aimless chaos.

There are two many issues for us to agree "the product of randomness" is much of an answer, I fear.
https://en.wikipedia.org/wiki/Randomness#In_biology
CF: https://phys.org/news/2022-01-evolution ... andom.html

https://en.wikipedia.org/wiki/Modern_sy ... h_century)

[machinesworking]

Everything in life that I've seen so far, from neurobiologists, psychology etc. leads me to the conclusion that the human brain is amazing, life is amazing, but it's not special.

So you're dismissing that consciousness is special? Actually no, in fact you only retreated to talk of the brain. All you've done is gainsay the central point with no consideration, providing only assertions with zero support.

I addressed this, to clarify again, Robert Sapolsky writes in Determined that what we think of as choice made by our own conscious will is not in fact a conscious choice. Synapses fire before the decision is made. In other words we are not in control really of our decisions, they're a mixture of DNA and environmental factors that led us to that decision at that moment. In general IMO the separate concept of consciousness is a way for us humans to pretend we're above other animals, or have some supernatural spiritual base, it's not based in anything like scientific consensus, and that's I think hard for people to accept.

[BONES]

So you're dismissing that consciousness is special?

Of course it's not special, there are more than 7 billion utter morons on this planet alone who can claim consciousness. It doesn't stop them from acting on their most primal instincts.

In sum: I cite Roger Penrose

Roger Penrose is a physicist, what the f**k does he know about it?

Darwin used the term "chance" to describe variations in organisms, which he later clarified meant a lack of understanding of the causes of variation, rather than true random, aimless chaos.

Yes, but he didn't know about DNA back then, which does explain the causes of variation.

[bermudagold]

so should using AI to clone a popular artists voice to increase probability of market penetration of your song be legal or illegal?...do you consider it immoral or unethical?...what if you can only make money by selling it, but you cant earn royalties on it, does that change your view?...what about if you cant monetize it at all but can publish it and disseminate it, does that change your view?

[Zeisner]

Well, can't say I'm disagreeing...

[machinesworking]

So you're dismissing that consciousness is special?

Of course it's not special, there are more than 7 billion utter morons on this planet alone who can claim consciousness. It doesn't stop them from acting on their most primal instincts.

There are 8,275,720,000 people approximately, so you're being kind for once, or hadn't checked the numbers recently.

The amount of credit we give ourselves for having an inner dialogue just proves what utter narcissists we are, I guess that goes back to base instincts? just funneled through supposedly complex ideas.

[vurt]

Well, can't say I'm disagreeing...

wait... chatgpt can do that? is that only a paid feature?

[BONES]

There are 8,275,720,000 people approximately, so you're being kind for once, or hadn't checked the numbers recently.

I don't think it's fair to consider babies morons until they are old enough to prove themselves, so I deliberately under-quoted on the numbers, then rounded down.

[jancivil]

The amount of credit we give ourselves for having an inner dialogue just proves what utter narcissists we are

The amount of credit you give yourself for tossing word salad like you're saying something is next-level. You have yet to make a single cogent argument on any point. This kind of shite in particular; this is an argument? This construction, a strawman: what amount of credit? Who is this we and what are you on about? I talk about a level of thought you won't consider and you go to these lengths to make it look shitty? That's all you. You need to sidestep what consciousness is entirely in order to believe some nonsense about a form of technology you don't appear to have a grasp of for shit in a type of childlike magical thinking; so you're at pains to reduce it to something you can dismiss with nary a thought: this time it's consciousness is but the capacity for an inner dialogue. This is deeply ignorant and ultra stupid. So you disagree with the most prominent thinkers that say "Consciousness is not computational". You might investigate that, but you know better. You're a fool.

You're so up your own ass, and apparently so enamoured of this fantasy about "AI" you're stuck, unreachable. This was never an exchange of ideas, it's me trying to enlighten the room and you thinking you know it all and posturing, so I feel no further need to hold back. Dunning-Kruger rules OK.

AGAIN: we can't locate consciousness. It is not a brain function, it is not physical.

At the quantum level potential becomes actual when the observer enters the picture, and the wave function collapses. This is beyond the grasp of classical physics entirely.
I think everyone willing to think should hear this: