I think we won't know when AI is conscious because we don't have a good definition of consciousness. Consciousness may not be an actual "thing" that exists.
@fourmajor I disagree to an extent. This is a "how many hairs make a beard" question. There is some liminal space, but your profile picture clearly has one and @freemo 's clearly lack one. There is a realm of existence on the edge of consciousness, but it is unstable. I propose that while there may be debate about what hit the historic marker of the first artificial consciousness, that it will be perfectly evident if and when such a thing comes into being.
🇳🇱
@antares @fourmajor With a beard though we have a pretty object sense of what a hair, we can identify a hair.. consciousnesses has nothing that is objectively measurable. Even I have no way in my own head, have any way to check to see if in conscious.
@freemo @fourmajor For all the well-worn philosophical reasons, yes, it is vexing to prove ones own consciousness. But, that was not the problem proposed. The question is to evaluate consciousness in another, which is a somewhat less abstract problem. I suspect it will be a value judgement (Is it noise or is it music?) but one that those of us reasonably convinced of our own consciousness will likely be able to come to some general agreement on if we don't let the skeptical philosophers get in the way.
As an aside I now want a short story a la @scalzi which is a first "person" account of an AI trying to prove it is the first conscious AI to a group of skeptical philosophers and failing. With the twist at the end being that the whole thing was an internal simulation and the AI concluding that it is not in fact conscious or self-aware.
@antares @freemo @fourmajor There are a lot of conflated ideas in this area that confuse people. The existence of consciousness, as in the incorrigible experiences of being, the qualia, the pure experience that make up ones existence, is something inaccessible to anything other than the being experiencing it and as thus the discussion of its existnece or nonexistence is pointless. What has a point is the question of whether something has personhood and thus should be treated as an agent with rights within our society. A side question is whether something is consious in the sense that neuroscientists talk about consiousness, which is really a discussion of a systems capacity to process information and integrate it into itself. The latter two topic deal with observable things while the first does not.
There is no conciousness from a neuroscientist perspective... AI can in some cases process and integrate more data than a human yet this would not qualify it as concious.
@freemo @antares @fourmajor I have read neuroscience papers where they talk about consciousness. In these papers, they are really implicitly assuming an identity between certain reported experiences and certain functions of the brain. These papers were focused on perception. Really, scientifically we can't talk about the "experiences" in and of themselves, but we can talk about the presence of the report which is evidence of some state of the system which we might call a "conscious" state by its ability to be reported. Additionally, these papers often used consciousness as a synonym for "awareness" which is why I talked about integrating information. In other words, neuroscience is the study of the relationship between the states of the brain and the behaviors of the agents in possession of the brain. We can say a subset of the information contained in the brain is information the agent is "aware" or "conscious" of by the agent's ability to report this information and to some degree their ability to otherwise act on it. This is the way I have seen "consciousness" used in some neuroscience papers.
Consciousness IMO is just an emergent phenomenon.. a trick the brain plays on itself. The evolutionary purpose is simply to make us want to avoid death at all cost.
@freemo @antares @fourmajor It's really the only thing in and of itself we have access to.
@freemo @mandlebro @antares @fourmajor
Federico Faggin (Olivetti, Intel, inventor of CMOS transistors, pioneer of x86 arch, founder of Synaptics, inventor of touchpads and touchscreens, early believer in neural networks as future of AI and even tried to implement them with anolog processing decades ago) has written a book on this that is skyrocketing in Italy (and I suspect it is available only in Italian).
The book is named "Irriducibile" (irreducible in Italian) and it explains, thanks to the Operational Probabilistic Theory developed by a Quantum Information professor from Italy, that it's impossible to reproduce consciousness artificially.
@post @freemo @antares @fourmajor Sounds like complete woo. Anything that connects the philosophy of consciousness to quantum theory is highly suspicious. It either has additional metaphysical assumptions , an additional dubious assumption about an identity between consciousness and some physical thing, or is complete nonsense.
@mandlebro @freemo @antares @fourmajor
Looking at your username it is ironic that the key concept in Faggin's reasoning is fractals.¹
Of course it sounds "woo" to you; do you think it is easy to summarise the content of a book on such a complex argument with a few sentences?
--
1: Check prof. Giuseppe Vitiello work too, here there is a video: https://www.youtube.com/watch?v=HqVULUI76L4
Here there is a talk by Faggin on his book (in Italian) organized by Vitiello and others: https://www.youtube.com/watch?v=koniUhyrMXM
@freemo @antares @fourmajor I feel all aspects of the mind can be plausibly explained by evolutionary pressures and biological and physical constraints, except for actual conscious experience. The thing I am experiencing right now as I write this, that I cannot share with anyone else, that I cannot even communicate about because it is uniquely mine and language requires commonality to form semantics. Some aspects of this thing can be understood by the aspects of the mind, but its existence and the innate nature of the primitive elements that constitute it are utterly beyond examination.