First, please read Bernie's excellent thread on AI.
-
Wasn't the OP about generative AI rather than AGI (artificial general intelligence; disembodied consciousness etc)?
Tech bros collapse the difference but it's a gulf, linked only by the letters being A and I.
OTH, Turing's original test assumed a disembodied humanity so the distinction I have made doesn't matter for your general point
-
So this idea that there can be an actual AI, that there can be an unembodied consciousness, is the most outrageous of all delusions.
Because there can be no real intelligence, no real consciousness, without finitude. And there can be no tiny glint of the real behind, beneath, at the edges of our magnificent imaginations, or the production of the infinitely elaborate symbolic world of language we are immersed in, without it being produced by a body that will one day stop being. 5/
@Remittancegirl well, you’re making a pile of assumptions there but I’d guess that we’d have real trouble relating to an intelligence that wasn’t embodied similarly to us. I don’t know what references we’d have in common.
You’re correct that the “mind piloting a meat robot” view is nonsensical dualism.
But all this is all orthogonal to the current conversation about LLMs, which aren’t intelligent or sentient at all.
-
OTH, Turing's original test assumed a disembodied humanity so the distinction I have made doesn't matter for your general point
While I have a tremendous affection for Turing, I've never accepted his 'test' as being proof of anything beyond our own desire to discern bunnies in cloud formations.
-
While I have a tremendous affection for Turing, I've never accepted his 'test' as being proof of anything beyond our own desire to discern bunnies in cloud formations.
Yes. The longer I have thought about it the more I see it as a test that presumes its own answer. Remove all evidence of what makes us human except symbolic interaction (language) then ask if we can be fooled under only that condition. Answer: of course.
-
@Remittancegirl thank you for this thread - an illuminating way of considering AI
@alstonvicar You are most welcome. It's something I think about a lot.
-
Yes. The longer I have thought about it the more I see it as a test that presumes its own answer. Remove all evidence of what makes us human except symbolic interaction (language) then ask if we can be fooled under only that condition. Answer: of course.
The better test: will a snail avoid discomfort? yes. Will an AI Turing test machine even be sentient? No.
-
The better test: will a snail avoid discomfort? yes. Will an AI Turing test machine even be sentient? No.
@craigduncan I love this encapsulation, because it plays out very elegantly.
Why does a snail avoid discomfort? What mechanism causes a snail to avoid discomfort?
Is it consciousness or instinct? Actually, it doesn't much matter which, because ultimately it is powered by an imperative to keep the body alive.
-
@lipservant But even if it became far more than that, it is still undead and parasitic.
-
@Remittancegirl well, you’re making a pile of assumptions there but I’d guess that we’d have real trouble relating to an intelligence that wasn’t embodied similarly to us. I don’t know what references we’d have in common.
You’re correct that the “mind piloting a meat robot” view is nonsensical dualism.
But all this is all orthogonal to the current conversation about LLMs, which aren’t intelligent or sentient at all.
@Colman Mind/body dualism is fallout from Christianity and earlier religions that posited an afterlife. Can't have an afterlife without some "essence" that survives bodily death! Which is thus problematic because it's both obviously bullshit but also hugely attractive to primates with a hard-wired terror of personal death (due to evolution selecting out strains that lacked that trait).
-
@craigduncan I love this encapsulation, because it plays out very elegantly.
Why does a snail avoid discomfort? What mechanism causes a snail to avoid discomfort?
Is it consciousness or instinct? Actually, it doesn't much matter which, because ultimately it is powered by an imperative to keep the body alive.
Yes, your thread is right on this, I just happen to have been thinking about that metaphor today (it fell out of a longer note to myself, for some ongoing writing)
-
Yes, your thread is right on this, I just happen to have been thinking about that metaphor today (it fell out of a longer note to myself, for some ongoing writing)
@craigduncan I look forward to it!
-
@Colman Mind/body dualism is fallout from Christianity and earlier religions that posited an afterlife. Can't have an afterlife without some "essence" that survives bodily death! Which is thus problematic because it's both obviously bullshit but also hugely attractive to primates with a hard-wired terror of personal death (due to evolution selecting out strains that lacked that trait).
@cstross Yeah, I was compressing a lot into that line.
I don’t know how to work out how people from pre-industrial times would have seen it. Did they think they were a soul controlling puppet? When did we decide the brain was the seat of reason?
-
@cstross Yeah, I was compressing a lot into that line.
I don’t know how to work out how people from pre-industrial times would have seen it. Did they think they were a soul controlling puppet? When did we decide the brain was the seat of reason?
@Colman Ancient Egyptian beliefs about the soul and the afterlife are *fascinating* (and bits of Xtianity came from them—the Cult of Isis and her participation in the resurrection of Osiris shows through the Virgin Mary, for example). In particular, there were EIGHT different "souls" associated with different aspects of one person: https://en.wikipedia.org/wiki/Ancient_Egyptian_conception_of_the_soul
-
While I have a tremendous affection for Turing, I've never accepted his 'test' as being proof of anything beyond our own desire to discern bunnies in cloud formations.
@Remittancegirl @craigduncan My take on this isn't that LLMs passed the Turing Test. It's that human beings failed it.
-
@cstross Yeah, I was compressing a lot into that line.
I don’t know how to work out how people from pre-industrial times would have seen it. Did they think they were a soul controlling puppet? When did we decide the brain was the seat of reason?
-
@Colman Ancient Egyptian beliefs about the soul and the afterlife are *fascinating* (and bits of Xtianity came from them—the Cult of Isis and her participation in the resurrection of Osiris shows through the Virgin Mary, for example). In particular, there were EIGHT different "souls" associated with different aspects of one person: https://en.wikipedia.org/wiki/Ancient_Egyptian_conception_of_the_soul
@cstross daoism only has seven as far as I understand it.
-
@Remittancegirl @craigduncan My take on this isn't that LLMs passed the Turing Test. It's that human beings failed it.
Yes. Humans have all the conscious agency and ability to be fooled. The test is relational but wouldn't exist without humans being the benchmark. Flexible language also allows us to say "this machine fooled person X..."
-
So this idea that there can be an actual AI, that there can be an unembodied consciousness, is the most outrageous of all delusions.
Because there can be no real intelligence, no real consciousness, without finitude. And there can be no tiny glint of the real behind, beneath, at the edges of our magnificent imaginations, or the production of the infinitely elaborate symbolic world of language we are immersed in, without it being produced by a body that will one day stop being. 5/
@Remittancegirl There are really two different things being discussed: first, creating an intelligence that sets its own agenda and purpose. Second, uploading our own agenda and purpose into a new physical architecture. The first gives rise to the obvious question of what purpose? Our own purpose is driven by the evolution via survival of our physical bodies. Even at our most enlightened we are slaves to our need to breathe, drink and expel toxins. 1/
-
In a wider sense we are concerned with the survival of others like us and we are looking to these speculative creations to help that to continue. But why would they be interested in that? There are lots of examples of different types of living organisms helping other and some where one type is exploited by another at great detriment to its own functionality. Are we looking to be a parasite of these intelligences or work in symbiosis? If the latter, what are we offering? 2/
And this is where the second thing comes in: uploading our minds. To fully shake off the physical limitations of our bodies, we need a structure capable of holding an intelligence with purpose then we embed our own purpose within it. Great, except our purpose outside of our body’s and other bodies’ needs is very unclear. We can speculate, and writers have, but would uploaded minds be benevolent overlords, ruthless tyrants or simply uninterested in other meat bound beings? 3/3
-
@Remittancegirl There are really two different things being discussed: first, creating an intelligence that sets its own agenda and purpose. Second, uploading our own agenda and purpose into a new physical architecture. The first gives rise to the obvious question of what purpose? Our own purpose is driven by the evolution via survival of our physical bodies. Even at our most enlightened we are slaves to our need to breathe, drink and expel toxins. 1/
In a wider sense we are concerned with the survival of others like us and we are looking to these speculative creations to help that to continue. But why would they be interested in that? There are lots of examples of different types of living organisms helping other and some where one type is exploited by another at great detriment to its own functionality. Are we looking to be a parasite of these intelligences or work in symbiosis? If the latter, what are we offering? 2/
