The robots are coming. Restructure the economy. Go. - The Bear and Badger

GurtTractor

SpaceGazelle wrote:

What do you mean 'not even wrong'?

Unfalsifiable. As I understand it he seems to believe or has the intuition that consciousness is more fundamental than physical spacetime. The observations about the fuzziness of our perception are very useful and important, but he's put the cart before the horse.

SpaceGazelle

It's more mathematical than you might think. If the laws of physics can come out of game theory then it might be testable. Many physicists are trying to prove spacetime doesn't exist at all and they're pretty much trying to show this mathematically in a similar vein to Hoffman. Different maths admittedly, but they might converge on similar answers or at least similar mathematical structures.

GurtTractor

Mathematics was one of the core disagreements between Bach and Hoffman in that podcast, you might want to check out the bit about Gödel at least. Bach is a strong advocate for constructivism and computationalism, basically for something to be true it has to be implementable. A non-continuous fundamental reality follows from this, which seems to be a big point of contention among scientists.

As for consciousness Hoffman is in the "hard problem" mysterianism camp, which is basically; 'because I can not or have not developed a hypothesis for how the mind works, consciousness must be unexplainable and/or beyond our current physics'. It seems to be usually based around a fundamental confusion about the nature of subjective reality, a failure to reconcile dualism with materialism.

hunk

So, creating (artificial) intelligence: totally possible
Possible side effect of ai developing some form of consciousness over time: impossible because we do not understand how our minds work and consciousness comes about , simulated or otherwise.

poprock

hunk wrote:

So, creating (artificial) intelligence: totally possible
Possible side effect of ai developing some form of consciousness over time: impossible because we do not understand how our minds work and consciousness comes about , simulated or otherwise.

I think you could add ‘Also possible: AI reaching a point where it fools us into perceiving it as conscious, even though it isn’t.’

hunk

Yes, it's getting better and better at the Turing test.
Still, that possible side effect uncertainty tho.......

SpaceGazelle

GurtTractor wrote:

Mathematics was one of the core disagreements between Bach and Hoffman in that podcast, you might want to check out the bit about Gödel at least. Bach is a strong advocate for constructivism and computationalism, basically for something to be true it has to be implementable. A non-continuous fundamental reality follows from this, which seems to be a big point of contention among scientists. As for consciousness Hoffman is in the "hard problem" mysterianism camp, which is basically; 'because I can not or have not developed a hypothesis for how the mind works, consciousness must be unexplainable and/or beyond our current physics'. It seems to be usually based around a fundamental confusion about the nature of subjective reality, a failure to reconcile dualism with materialism.

I'll watch it.

SpaceGazelle

hunk wrote:

Yes, it's getting better and better at the Turing test. Still, that possible side effect uncertainty tho.......

The Bell understanding test Hossenfelder uses is nice, for now. I just asked Gemini and it got it wrong because the internet usually gets it wrong, so we can see it's just mimicking. It required a couple of follow up questions to check it got it wrong though.

The whole thing is so interesting. I mean, we know LLMs work by choosing the next word based on probability, in turn based on the question and what the internet says, which should seem obvious to us it's not sentient. But what do we do? Have a thought and then use the best words to try and get the point across.

We can come up with original thoughts like general relativity but AI can come up with new images. Can AI come up with new sentences? Yes, it can and does. Really we need to know the science of information better. Could we come up with new scientific theories just from language? Is the data buried in there? To add to the confusion the new AIs are multi-modal so they're using LLMs with other types of AI so they can combine bits of different ML learning.

Certainly as AI gets better and we use it more in our daily lives people are going to believe it's sentient, but it's going to be very hard to prove either way. We could just go full Blade Runner and say robots don't have the same rights but we will also give them more power in shaping society. It's a fucking minefield. I don't personally believe AI will get sentient for a long time but if I can't prove it isn't then that's just another belief isn't it?

SpaceGazelle

I say that and I don't even know what sentience is.

hunk

Exactly my point.
We're learning to run and fly but don't know how we got there and there's no landing gear in case of emergencies.

What is sentience and how does it come about? Is a frog or a fish sentient? How about a cat or a dog? A chimpansee or any great ape for that matter? Can we replicate the behaviour of simple organisms in AI algorithms? I'm willing to bet we can; it's just a matter of time. We'll likely be able to simulate human development of the mind too at some point. We're so caught up with the idea that we humans are special; our sentience given by god. It's our ego that's the problem and the fact that we're running head first into this AI thing blindfolded.

SpaceGazelle

GurtTractor wrote:

Mathematics was one of the core disagreements between Bach and Hoffman in that podcast, you might want to check out the bit about Gödel at least. Bach is a strong advocate for constructivism and computationalism, basically for something to be true it has to be implementable. A non-continuous fundamental reality follows from this, which seems to be a big point of contention among scientists. As for consciousness Hoffman is in the "hard problem" mysterianism camp, which is basically; 'because I can not or have not developed a hypothesis for how the mind works, consciousness must be unexplainable and/or beyond our current physics'. It seems to be usually based around a fundamental confusion about the nature of subjective reality, a failure to reconcile dualism with materialism.

The problem Bach has, being a computer scientist, is that the maths of infinities hasn't been fully fleshed out. You simply can't rule out stateless mathematics yet. At least with Hoffman he's saying we just don't know. Bach also tries to rule out continuous spacetime - which is fair enough, by using some notion of spacetime, which isn't.

We need more collaboration between computer scientists, neuroscientists, mathematicians and physicists to get through this, if we can.

SpaceGazelle

This may or may not have relavance. -1/12 comes up an awful lot in QM.

SpaceGazelle

Or at least when you just stick -1/12 in, the equations seem to work out and be testable.

SpaceGazelle

Just watched another vid on Bach and again he veers dangerously into religious psycho-babble.

GurtTractor

That'll be you not understanding an analogy he's making or the context. Happy to discuss it more if there's a specific bit you're referring to.

It's ironic that you describe it that way when Hoffman is basically using a lot of "well we just don't know yet" to justify his ideas, something very familiar to theological and psuedoscientific discourse. He stands on shaky ground and you can tell he knows it in that podcast.

SpaceGazelle wrote:

The problem Bach has, being a computer scientist, is that the maths of infinities hasn't been fully fleshed out. You simply can't rule out stateless mathematics yet.

Here is Bach saying the same things in a different way which might help -

The point he makes is that the maths of infinities cannot be 'fleshed out', because if a language contains contradictions it cannot be used to make proofs about truth. If you use a stateless language with infinities you can reason about things that cannot exist in reality, which can potentially be a problem for physicists if they make theories that rely on it. Classical mathematics offers useful abstractions and assumptions but we shouldn't conflate it with a description of reality.

Infinity and zero are everywhere in physics. Even in seemingly innocent things like space, or space-time. The moment you write down the mathematics for space, you assume there are no gaps in it. You assume it’s a perfectly smooth continuum, made of infinitely many infinitely small points.

Mathematically, that’s a convenient assumption because it’s easy to work with. And it seems to be working just fine. That’s why most physicists do not worry all that much about it. They just use infinity as a useful mathematical tool.

But maybe using infinity and zero in physics brings in mistakes because these assumptions are not only not scientifically justified, they are not scientifically justifiable. And this may play a role in our understanding of the cosmos or quantum mechanics. This is why some physicists, like George Ellis, Tim Palmer, and Nicolas Gisin have argued that we should be formulating physics without using infinities or infinitely precise numbers.

http://backreaction.blogspot.com/2020/12/is-infinity-real_5.html

SpaceGazelle

The point of the "well we just don't know yet" is led by experiment, and Bach doesn't seem to care about that. He just babbles as far as I can tell. There are no contradictions in maths despite what Bach says. Feel free to point out a specific example.

GurtTractor

Do you believe that infinities can exist?

SpaceGazelle

Who gives a shit? Point out what is contradictory about mathematics and we can talk. With examples that heads up, you won't be able to provide.

hunk

Eehm, a circle is an infinite....

SpaceGazelle

Some circles are infinite. It's still a bullshit argument.

hunk

Agreed, maths is kindof the best descriptive model we have of nature and physics. Because the human mind has trouble fathoming a concept like infinite doesn't change that fact.

SpaceGazelle

Bach is still a mental. Total idiot unless he can prove his mad theories.

SpaceGazelle

GurtTractor wrote:

That'll be you not understanding an analogy he's making or the context. Happy to discuss it more if there's a specific bit you're referring to. It's ironic that you describe it that way when Hoffman is basically using a lot of "well we just don't know yet" to justify his ideas, something very familiar to theological and psuedoscientific discourse. He stands on shaky ground and you can tell he knows it in that podcast.

SpaceGazelle wrote:

The problem Bach has, being a computer scientist, is that the maths of infinities hasn't been fully fleshed out. You simply can't rule out stateless mathematics yet.

Here is Bach saying the same things in a different way which might help - The point he makes is that the maths of infinities cannot be 'fleshed out', because if a language contains contradictions it cannot be used to make proofs about truth. If you use a stateless language with infinities you can reason about things that cannot exist in reality, which can potentially be a problem for physicists if they make theories that rely on it. Classical mathematics offers useful abstractions and assumptions but we shouldn't conflate it with a description of reality.

Infinity and zero are everywhere in physics. Even in seemingly innocent things like space, or space-time. The moment you write down the mathematics for space, you assume there are no gaps in it. You assume it’s a perfectly smooth continuum, made of infinitely many infinitely small points. Mathematically, that’s a convenient assumption because it’s easy to work with. And it seems to be working just fine. That’s why most physicists do not worry all that much about it. They just use infinity as a useful mathematical tool. But maybe using infinity and zero in physics brings in mistakes because these assumptions are not only not scientifically justified, they are not scientifically justifiable. And this may play a role in our understanding of the cosmos or quantum mechanics. This is why some physicists, like George Ellis, Tim Palmer, and Nicolas Gisin have argued that we should be formulating physics without using infinities or infinitely precise numbers.

http://backreaction.blogspot.com/2020/12/is-infinity-real_5.html

Do you think that post is somehow good? Like it's true?

GurtTractor

SpaceGazelle wrote:

Point out what is contradictory about mathematics and we can talk.

Might be time to dust off my old Let's have an argument thread

SpaceGazelle

Yes, but the Universe doesn't break does it? Bertrand Russel had a similar problem in trying to define numbers not in terms of themselves. We could regard "This statement is false" as an incomplete statement.

SpaceGazelle

I suspect the problem is it assumes a falsehood where there might be none. For a start there's no equals sign in that statement.

SpaceGazelle

Ok, I've had a think about this and we need to frame it mathematically.

"This statement is false" = False.

Well that isn't true is it? The LHS does not equal the RHS. The concept of a statement doesn't even equal the concept of False.

monkey

I preferred this thread when it was about robots.

SpaceGazelle

But the Universe might be a robot so it's fine.

SpaceGazelle

Unless it runs out of batteries.

Howdy, Stranger!

Categories