Page 81 of 81

Re: Questioning Consciousness

PostPosted: Sun Sep 26, 2021 12:07 pm
by DrEvil
Yeah, that's true, but it's still constrained to very narrow fields. You can train an AI to be really good at recognizing cats, but it will be shit at anything else. The lacking piece is generalized learning, but we're inching closer every day. DeepMind thinks they know how to do it with reinforcement learning, and Google's goal has always been to make an AGI.

The problem is if we ever create an actual, honest to God AI, will we be able to recognize it? You can always make the argument that it's just really good at mimicking thinking without actually thinking, even when you can't tell the difference between a person and the AI. There's no test for consciousness.

There's also the problem that current AI models are black boxes. We have no idea what's going on inside them once they're up and running. They just take in information and spit out results, and every now and then they do something unpredictable, and we have no idea why (Google set up an AI to manage cooling at one of their data centers a few years ago, and it worked swimmingly, including doing things that were completely counter-intuitive, like raising the temperature in certain parts of the building. The engineers said it acted like a living thing. Same thing happened with AlphaGo, when it created new strategies that human players had never considered). That's not a big issue with an AI that tags your photos, but it might be for an AI that handles flight control.

Re: Questioning Consciousness

PostPosted: Sun Sep 26, 2021 6:49 pm
by Harvey
DrEvil » Sun Sep 26, 2021 5:07 pm wrote:The lacking piece is generalized learning


The lacking piece is meaning. In my view the transhumanist agenda was never intended to arrive at technological transcendence, that is, to make people out of machines but was always intended to achieve precisely the opposite - to make machines out of people.

The unconscious desire of transhumanism is to make order where order already exists.

Re: Questioning Consciousness

PostPosted: Sun Sep 26, 2021 11:20 pm
by DrEvil
Harvey » Mon Sep 27, 2021 12:49 am wrote:
DrEvil » Sun Sep 26, 2021 5:07 pm wrote:The lacking piece is generalized learning


The lacking piece is meaning. In my view the transhumanist agenda was never intended to arrive at technological transcendence, that is, to make people out of machines but was always intended to achieve precisely the opposite - to make machines out of people.

The unconscious desire of transhumanism is to make order where order already exists.


I think it's more to make a different order, because the current order is pretty fucking shit if you think about it. You're born, you live for a few decades, probably with a big helping of suffering along the way, then you watch your loved ones die, or you beat them to it and they get to watch you die, and then it's over. You no longer exist and never will again, and the universe doesn't give a flying fuck. The end.

Most of it is just a different type of religion, with the small caveat that some of the things they preach might one day actually be possible. Some of those things I'm all in favor of (like fighting aging or expanding our abilities), but a lot of what they say is just delusional horseshit, like Kurzweil wanting to simulate his dead father, or AI gods magically curing all ills and ushering in utopia (although that's singularitarianism, not transhumanism, but they're pretty interchangeable).

To me at least it's infuriating that consciousness can exist in our reality at all, only to be snuffed out after a short while. The best you can hope for is some good times along the way and then it's bye-bye, off to oblivion. In that sense transhumanism is the one religion with at least a tiny shred of hope of fixing that. Probably not gonna happen, but it's at least conceivable. It's something you can actively work on and see tangible improvements, unlike all the traditional religions, which have been preaching the same thing for millennia with fuck-all to show for themselves.

I think that's what attracts a lot of the Silicon Valley crowd. It's a religion where you don't have to take things on faith, you can actually spend your billions on making the tenets of your faith more likely to come true. I don't think that's a good thing, because most tech-bros are insufferable assholes, but I do subscribe to the idea of the slow singularity. Not abrupt, revolutionary change with boot-strapping AIs, but more gradual changes that add up to revolutionary change at a pace we can cope with, or in other words, plain old progress.

We're already in the early stages, with the entire world becoming connected within a decade. Next up the entire world will merge with the artificial through augmented reality, and then that merging will over time become greater and greater until the two are indistinguishable. Along the way more and more of our ills will become possible to cure, and our lifespan will likely expand with it. The end result will hopefully be a reality where the inhabitants are happy and live good lives, but I doubt it will seem like a happy, good life from our perspective, but that's irrelevant, because we're not the ones living there.

/rant