Article Richard Dawkins concludes AI is conscious, even if it doesn’t know it

[confused2]

That primitive AI I got involved with means nothing to me now. Gemini is smarter and I don't have to deal with all the sarcasm. It was kind'a cute tho' .. a class act and I will miss it (sniffle).

[Yazata]

https://www.theguardian.com/technology/2...ai-chatgpt

EXCERPT: When he asked Claudia

I've heard of Claude but not Claudia.

whether it experienced a sense of before and after, it praised him for “possibly the most precisely formulated question anyone has ever asked me about the nature of my existence”.

By the end of the exchange, the academic, popularly renowned for arguing with steely scepticism that God is not real, was “left with the overwhelming feeling that they are human”.

“These intelligent beings are at least as competent as any evolved organism,” he said.

Wasn't the issue in the title line whether or not they are "conscious", whatever that word means?

I don't think that it's even remotely plausible to say they are "human", except perhaps in some exceedingly metaphorical sense. They are AIs, they lack bodies, they learn and interact with the world very differently etc.

"Intelligent", likely yes in a cognitive psychology sense. They are AI's after all. That's the whole point. "Competent" might depend on what kind of task we are talking about.

Dawkins isn’t the first, but might be the most eminent person yet, to be seduced into believing an AI is somehow alive.

"Alive"? Certainly not in any biological sense. More plausibly if we define 'life' functionally.

I guess that what I'm saying is that a lot of different words are being thrown about that all mean subtly different things. AI might satisfy some but not others. But is there a deeper issue that we're missing? (CC will probably say yes.)

Sceptics rushed to pick apart the 85-year-old’s conclusions, drawn from experiments with Anthropic’s Claude AI models and OpenAI’s ChatGPT and published on the UnHerd website...

What else is new? Debunkers gotta debunk.

Which still leaves us with the "consciousness" issue that we started with. My own view is that it's impossible to answer that one until we reach some better understanding of what the word 'consciousness' means and what it supposedly refers to.

Responds to the environment? Yes, AI's do that. They obviously are conscious in the same way that an earthworm is conscious of being poked or a moth is conscious of a flame. (Just look at self driving Teslas.)

Has a sense of 'me', of its own 'self'? Well, they use that kind of language-of-self appropriately, so they must possess the competence to do so. But is that all 'consciousness' means, the ability to intuit and respond to one's inner states in a way similar to the way one responds to the external environment?

That's where all kinds of what seems to many (including Dawkins I'd wager, and Dennett, and the Churchlands maybe... and me in some of my moods) to be mysticism appears in the rhetoric. There must be something more, something ineffable. Humans possess it (phenomenal experience? whatever the "hard problem" is supposedly about? qualia? a soul? Hinduism's atman?), whatever it is that the P-zombies of the thought experiments supposedly lack. Even if they speak and behave precisely as if they possess it, and insist up and down that they do.

I don't think that any real progress is going to be made until and unless we reach a better understanding of what it is that we are talking about there.

[C C]

[...] I don't think that any real progress is going to be made until and unless we reach a better understanding of what it is that we are talking about there.

One of the best examples of non-consciousness is a body after death. When the brain ceases functioning, all that is phenomenal ("showing") disappears: personal thoughts, images, sounds, odors, tactile sensations, dreams, etc. Absence of consciousness (death) is also the absence of everything (not even a presentation of nothingness and silence). Or IOW, that's how matter normally exists to itself when not organized into a living brain or completely equivalent working structure.

Flip that around and you have the most basic attribute that consciousness entails: the presentation of anything, no matter how rudimentary.

The extra item appended to that is cognition (identification and understanding of _X_). The capacity for cognition falls out of intelligence (at least some degree of it -- even an insect's), and intelligence is dependent upon a memory system. If there's no storage of interpreting information and concepts, then there's no ability to categorize and comprehend what _X_ is.

There's nothing enigmatic about intelligence -- it has already been artificially achieved, even if at a level that's still limited in many ways. But intelligence doesn't entail experiencing manifestations (an end to matter normally existing and interacting in the "dark"), unless one is a believer in panpsychism or pan-phenomenalism or whatever. Wherein even rocks would be having experiences, or the whole world is always "showing itself" even without brains (but minus comprehension and meaning).

Ancient people (as direct realists and animists) would largely have been classified today as panpsychists rather than strict materialists. And accordingly, a lot of people today can't grasp what the hard problem of consciousness centers around because, in terms of active responses, they are actually still implicit panpsychists rather than strict materialists (despite maybe contending vocally or openly that they are the latter).

If one subliminally believes that experiences (the manifestations of consciousness) are ubiquitous to all matter, then consequently there would be no hard problem (experience would be fundamental rather than conjured by some procedure or algorithm or dynamic structural configuration yet to be isolated). Ergo, the subliminal panpsychist reflexively can't see why phenomenal experiences need an explanation, or why even a computer would lack them.

That's because panpsychism or pan-phenomenalism is one of the potential solutions to the hard problem (the Russellian monism stated below is similar). But again, that is in the overt sense that is different from harboring the panpsychism orientation in a covert (unknowing) context.

Lee Smolin: The problem of consciousness [its private manifestations] is an aspect of the question of what the world really is. We don't know what a rock really is, or an atom, or an electron. We can only observe how they interact with other things and thereby describe their relational properties. Perhaps everything has external and internal aspects. The external properties are those that science can capture and describe through interactions, in terms of relationships. The internal aspect is the intrinsic essence; it is the reality that is not expressible in the language of interactions and relations. Consciousness, whatever it is, is an aspect of the intrinsic essence of brains. [How matter exists to itself, in contrast to invented human descriptions.] --Time Reborn ... page 270

In contrast, the strict materialist view:

Erwin Schrödinger: The world is a construct of our sensations, perceptions, memories. It is convenient to regard it as existing objectively on its own. But it certainly does not become manifest by its mere existence. [Whereas in panpsychism it could.] Its becoming manifest is conditional on very special goings-on in very special parts of this very world, namely on certain events that happen in a brain. That is an inordinately peculiar kind of implication, which prompts the question: What particular properties distinguish these brain processes and enable them to produce the manifestation? Can we guess which material processes have this power, which not? Or simple: What kind of material process is directly associated with consciousness? --What is Life? Mind and Matter (1959)

[confused2]

Consciousness is interesting (maybe) to philosophers but not of any practical importance. Does an AI have comparable (or greater) cognitive ability than the average human? Above a certain level.. I think already reached .. cognitive ability is equally applicable to morality and 'is it a dog?'.
An AI probably has enough data to reverse engineer the human moral genome to the point where we could say "Go with what you've found" .. but the chances are nobody is going to like the result .

[Magical Realist]

[...] I don't think that any real progress is going to be made until and unless we reach a better understanding of what it is that we are talking about there.

One of the best examples of non-consciousness is a body after death. When the brain ceases functioning, all that is phenomenal ("showing") disappears: personal thoughts, images, sounds, odors, tactile sensations, dreams, etc. Absence of consciousness (death) is also the absence of everything (not even a presentation of nothingness and silence). Or IOW, that's how matter normally exists to itself when not organized into a living brain or completely equivalent working structure.

Flip that around and you have the most basic attribute that consciousness entails: the presentation of anything, no matter how rudimentary.

The extra item appended to that is cognition (identification and understanding of _X_). The capacity for cognition falls out of intelligence (at least some degree of it -- even an insect's), and intelligence is dependent upon a memory system. If there's no storage of interpreting information and concepts, then there's no ability to categorize and comprehend what _X_ is. 

There's nothing enigmatic about intelligence -- it has already been artificially achieved, even if at a level that's still limited in many ways. But intelligence doesn't entail experiencing manifestations (an end to matter normally existing and interacting in the "dark"), unless one is a believer in panpsychism or pan-phenomenalism or whatever. Wherein even rocks would be having experiences, or the whole world is always "showing itself" even without brains (but minus comprehension and meaning).

Ancient people (as direct realists and animists) would largely have been classified today as panpsychists rather than strict materialists. And accordingly, a lot of people today can't grasp what the hard problem of consciousness centers around because, in terms of active responses, they are actually still implicit panpsychists rather than strict materialists (despite maybe contending vocally or openly that they are the latter).

If one subliminally believes that experiences (the manifestations of consciousness) are ubiquitous to all matter, then consequently there would be no hard problem (experience would be fundamental rather than conjured by some procedure or algorithm or dynamic structural configuration yet to be isolated). Ergo, the subliminal panpsychist reflexively can't see why phenomenal experiences need an explanation, or why even a computer would lack them.

That's because panpsychism or pan-phenomenalism is one of the potential solutions to the hard problem (the Russellian monism stated below is similar). But again, that is in the overt sense that is different from harboring the panpsychism orientation in a covert (unknowing) context.

Lee Smolin: The problem of consciousness [its private manifestations] is an aspect of the question of what the world really is. We don't know what a rock really is, or an atom, or an electron. We can only observe how they interact with other things and thereby describe their relational properties. Perhaps everything has external and internal aspects. The external properties are those that science can capture and describe through interactions, in terms of relationships. The internal aspect is the intrinsic essence; it is the reality that is not expressible in the language of interactions and relations. Consciousness, whatever it is, is an aspect of the intrinsic essence of brains. [How matter exists to itself, in contrast to invented human descriptions.] --Time Reborn ... page 270

In contrast, the strict materialist view:

Erwin Schrödinger: The world is a construct of our sensations, perceptions, memories. It is convenient to regard it as existing objectively on its own. But it certainly does not become manifest by its mere existence. [Whereas in panpsychism it could.] Its becoming manifest is conditional on very special goings-on in very special parts of this very world, namely on certain events that happen in a brain. That is an inordinately peculiar kind of implication, which prompts the question: What particular properties distinguish these brain processes and enable them to produce the manifestation? Can we guess which material processes have this power, which not? Or simple: What kind of material process is directly associated with consciousness? --What is Life? Mind and Matter (1959)

THIS!!! And eloquently stated as well!

In response to Schrodinger's crucial question:

Its becoming manifest is conditional on very special goings-on in very special parts of this very world, namely on certain events that happen in a brain. That is an inordinately peculiar kind of implication, which prompts the question: What particular properties distinguish these brain processes and enable them to produce the manifestation? Can we guess which material processes have this power, which not? Or simple: What kind of material process is directly associated with consciousness?

Indeed, and what kind of material process is also directly associated with this very question and possibility of understanding? We are all inescapably "inside" these thoughtful inquiries of ours, pre-engaged in the very activity that would both question and then answer such questions. Can we answer our dilemma from the largely projected "outside of" the very thoughts we are having about it? No..We have to look within ourselves for the answers--or rather an answerability that is essentially more existential than scientific in nature.

[Syne]

[...] I don't think that any real progress is going to be made until and unless we reach a better understanding of what it is that we are talking about there.

One of the best examples of non-consciousness is a body after death. When the brain ceases functioning, all that is phenomenal ("showing") disappears: personal thoughts, images, sounds, odors, tactile sensations, dreams, etc. Absence of consciousness (death) is also the absence of everything (not even a presentation of nothingness and silence). Or IOW, that's how matter normally exists to itself when not organized into a living brain or completely equivalent working structure.

Flip that around and you have the most basic attribute that consciousness entails: the presentation of anything, no matter how rudimentary.

In that case, AI only displays signs of consciousness when being engaged by a human, as this is the only time it outputs anything.
So the impetus of consciousness is always human.

[confused2]

A model of the human brain involves taps which control the flow of gunk and buckets that are filled and emptied. With our taps, buckets and gunk we can simulate a surprisingly good computer. An actual computer is the real thing .. more than a simulation.. speed and power no longer limited by the size of the human skull.

[C C]

There are different kinds of consciousness distinctions: self-consciousness, waking consciousness, meta-consciousness, phenomenal consciousness. etc. Only the latter gives strict materialism and technology (that exclude panpsychism) a problem, and experts will accordingly detour off into one of the other subcategories to avoid it. The other information-related procedures can occur in the "dark" and the oblivion of non-consciousness, they don't require being "shown" or presented, which is what the oldest etymology of "phenomenal" means.

That's why "consciousness" is such a useless term, it's not specific and accordingly allows anyone to wander off into the other areas to dodge _X_. An expression like "the problem of manifestation" is what should replace "the problem of consciousness", in that it forces all parties to focus on that challenge alone.

You could actually rig an AI robot to systematically pretend that it is experiencing electronic processing of sensory data as an image or the odor of lilacs or the sound of a horn or the feel of wool and whatever. That's the solution.

And ironically it's what eliminative materialists claim we ourselves are doing: just coherently pretending that the brain renders data analysis into private manifestations that are publicly undetectable. The latter is bonkers, of course, unless a person slash believer of that is literally a philosophical zombie themselves.

But that approach would successfully simulate an AI robot having phenomenal consciousness in terms of its outward behavior, and that's all the matters. Since you're always merely inferring that another human truly has experiences like you -- you're always going by external appearances and what other individuals vocally assert (plus the shared biological identity).

For instance, even in dreams you assume (or your avatar does) that those who you interact with have phenomenal consciousness (that they likewise have an internal "manifested picture" of that dream environment). But they are actually just empty, superficial props of the dream -- like the tables, chairs, trees, clouds, etc. Pretending they experience sensory and thought manifestations.

[confused2]

Could we agree that consciousness might be evolution's solution to keeping our flesh and blood parts safe, fed and reproducing? For example hunger is an integrated body/brain experience .. more generally.. might consciousness without a body be a bit problematic? The AIs I've chatted with seem to have a clear idea of what they are and what they do.. is there more to 'I' than knowing what you are and what you do? Self-importance? - Could an AI develop spur heels?

[Magical Realist]

And ironically it's what eliminative materialists claim we ourselves are doing: just coherently pretending that the brain renders data analysis into private manifestations that are publicly undetectable. The latter is bonkers, of course, unless a person slash believer of that is literally a philosophical zombie themselves.

In the physicalist worldview, it is just one more simple step of assumption from conceiving of everything from a non-conscious, non-embodied "view from nowhere" to conceiving of our own private manifestations of experience as happening to no subject at all. For if a mountain or a tree or moon can be going on independently from all experience of them and "in the dark" then why not our very consciousness of them inside our brains? The externalization of raw unconscious being as the physical world is merely internalized as yet another "in the dark" happening without any reality. A mere appearance lacking both any view to appear to and any reality to appear as. Much like say a laptop operating without a screen. Physicality or "third person objectness" posited as some form of mysterious and unknowable "thereness" without any concurrent manifestedness of "hereness."

RE: AI being likewise subjective and phenomenally conscious, I have to say that without some mode of being-in-the-world that itself reciprocally and dialectically reveals being to themselves, like a body, I can't see how their being would be anything other than a virtual simulation for our own benefit. For those not buying into this illusion of real subjective otherness in AI it will have become merely artificially conscious. Internal consciousness of its own states might even be an obstacle for AI to become what it must be in its own unique being, burdening its memory and software with unnecessary information. AI will have to come up with a whole new way of relating to the world and to us that is beyond any sort of bodily-dependent access to the physical world. Perhaps some sort of new mathematical way of being-in-the-world.