AI becoming sentient

Interesting, but I strongly disagree with the conclusion that the only thing that matters is how humans feel about it.

Possibly. The brain might not be just a chemical process - it could be functioning as a quantum interface to mystical astral beings from Energon 6 who are just puppeting us because they like roleplaying as meat beings.

The whole quantum brain idea has been floating around for a while and is still being studied: www.quantamagazine.org/a-new-spin-on-the-quantum-brain-20161102/

If the human illusion of consciousness is due to a quantum effect then a classical computer, no matter how complex, would not be able to recreate it. But maybe a quantum computer could? We'll have those soon enough, yeah?

Part of the problem is that we don't actually understand how humans are particular or why that is. Neuroscience keeps advancing at a rate of knots, regularly overturning older perceived orthodoxies. So that human brains create emergent behaviour we class as consciousness is one thing, what the actual definition of consciousness is or how the brain creates it is still basically a big mystery. So exactly copying / taking lessons from that directly is still quite an inaccurate process, that if you were keeping your good scientist hat on you wouldn't go making predictions.

Obviously you'd put it helpfully on a spectrum including animals like apes at one end, my cat in the middle and prawns or what not at the other. Though try and find a group of people who all agree on what that actually means ethically and you'll also get stumped.

But that also gives us a handy chart of body shapes to hazard a guess from. Disembodied things would be harder to pin down, however considering they can they be copied at will and run in multiple places, demonstrating fear of being switched off is actually a lot odder. I still suspect when someone finally cracks this they may initially be embodied in some sense as constructing experiences and therefore narrative about it seems important to being human, even 2 year olds can manage to make up shit stories and start to lie. Searching a massive dataset of things human's say that would fit isn't really quite the same thing as experiencing it. There's the old philosophy thought experiment of imagining a child who's grown up in the dark; they've had colour explained to them, the experience of colour, the science of light etc. All of it. Is that at all the same as stepping outside and actually experiencing it for themselves?

Anyhow the whole thing reminds me of the Chinese Room thought experiment.

I mean everything we see could be the hallucinations of a dying brain that spontaneously formed in the vacuum of space. Given an infinite timescale this is fundamentally possible.

Or even if they are.

Cognitively, we have the a greater ability for abstraction than other species, but that exists on a spectrum. Other primates learn how to use tools and can learn from each other. There seems to be a neurophysiological explanation for empathy (mirroring) that is evident across species, which implies that the concept of self and others and how they relate is more basic than we suspect.

Yeah Roger Penrose was a complete chancer starting on that quantum bullshit.




Yeah it is basically impossible to analyse any human intelligence separate from their bodies and state of tiredness/hunger/horniness etc.

The post-hoc narrative thing is a bit out of date, from the last thing I read but I'm no neuroscientist. As it turns out a bunch of those experiments about people rationalising reactions afterwards only happens under certain circumstances (handily lab based ones). Though it did always seem far too neat that there was basically 1-3 processes going on, reacting to shit, communicating about said shit and then also woolily recasting that as narrative. It'd make everyone's life a lot easier if that was the only thing our brains did.

Which is why I think this google AI thing is basically a fudge.

A lot of what we experience as consciousness is at very least heavily informed (if not, imho, entirely dependent on) fairly mundane biological process. There's just a lot of them that all interact. AI as above is a simulation. The secret sauce is how much is genuinely emergent and how much is pre-ordained according to how the neural networks underlying it as created. Is there a specific neural net to extract emotional tone from processing of text, or is that something that emerged spontaneously which the engineers have no explanation for? I'd imagine at least some of the former.

I think a true AI would be able to communicate concepts to us which are entirely novel to us which are the result of it being based on a different substrate. Anything else is just an impersonation.

Novel concepts, not incomprehensible.

I'd expect it to be different, but be able to describe how they're *different*. Not dwell on how they're the same as us.

Biological processes are not infallible though. Multiple imperfect interacting feedback loops are not going to act entirely deterministically and are going to produce outcomes that are more probabilistic and adaptive rather than deterministic. You don't need to introduce quantum mechanics (which...er....is a mathematical model anyway: just a probabilistic one) to avoid determinism.

To a man with a hammer, a lot of things start to look like nails.

One thing's for sure, whoever created the simulation is having a good laugh at the signs of sentience we're showing. "Hey look, they're debating what it is to be human, bless".

Well, it does involve Gödel’s incompleteness theorem, there’s things that we know to be true, but can’t be calculated or proven. We have no trouble imagining those things because we can place ourselves outside the system. Not saying I agree, but there’s more there besides we don’t understand consciousness, we don’t understand quantum physics, so they’re probably related.

Not quite, the incompleteness theorem would say that our current language of science (rather than maths, but that makes the whole invocation slippier) can't describe this. We will therefore need to add new structure to that language to describe it, which will beg more questions which will be incomplete within the framework of that language, which will need extension. Etc. ad infinitum.

So invoking it in this context is more akin to what I was saying, we barely understand consciousness at the moment anyway, so we just don't know right now. E.g. if its unknowable or properly describable currently one should pass over it in silence to borrow a Wittgenstein quote.

Not. Hand wave, hand wave, oooooo QUANTUM PHYSICS.

I have 3. Combined mass of approx 150 kg.

The Go AI kind of did that. It played sequences of moves that humans thought were completely bollocks, but later on turned out to have been very advantageous. Sure, it's not general intelligence, but it still learned moves by playing against itself that no other human had discovered before.