In a discussion about consciousness on Mastodon, I noted that consciousness is a state, not a quality. That’s important, when we talk about whether or not entities (whether they be human-shaped or AI-shaped) as having "consciousness" and "self-awareness".
Mikeydangerous asked a really important question that’s been bothering me for a while when looking at myself:
… how would you define the difference between understanding and pattern prediction in relation to consciousness? That’s how AI generates the intelligent-sounding platitudes it does. It sees the context of those sentiments in enormous sets of data and builds it from there with no real understanding or creativity. At best, it’s scaled-beyond-human capability pattern matching. At worst, it’s scaled-beyond-human capability plagiarism.
Because that’s where Descartes falls down, isn’t it?
Right, quick recap. Descartes, while writing to convince the Roman Catholic Church that science isn’t inherently heresy, tries to find some kind of rock-solid truth, some certainty that he can build from. It’s from this that we get "I think, therefore I am;" Descartes states that since there is a point of view, that point of view must exist.
He then immediately jumps straight to "my senses must be accurate, because God wouldn’t be mean like that," which we now know is absolutely not the case at all. Our senses are fickle, easily fooled, and prone to error. That’s important because while there are visual tests (Necker Cube, the mirror test, the one I suggested) that may correlate with being conscious, they could also simply be artifacts due to the makeup of our sensory "hardware" and the ways our brains process visual information rather than being a sign of consciousness in ourselves and others. So let’s ignore that, and go back to Descarte’s central thesis: I think, therefore I know that I am.
Given what we know about consciousness, as MD points out, how do you tell the difference between being conscious and being a sophisticated pattern-matching machine? How do you know that any entity — including yourself — is not a philosophical zombie?
Before someone says it, no, art is not some safe haven of consciousness. We are only able to "tell" the difference between AI generated "art" and the work of inspired humans only by the errors it makes or its repetitive nature. Once again, it’s a processing and fidelity issue, not some special quality only possessed by consciousness. (Wait with your objections, folks.)
That idea becomes a lot more immediate and real when you’ve had the fact that there are non-conscious (semi?) autonomous routines that are capable of moving your body around and interacting with the environment without bothering to inform your consciousness. {1} And unlike Descartes, we know that our senses are inaccurate and betray us — not only in terms of effects too small for us to typically notice (such as the relativistic effects of when you tap your toes to a catchy song), but because even if we accept that emotions are something more than just biochemical reactions, those biochemical reactions do occur.
But that’s where our philosophical zombies come to the rescue, in a way. Or rather, testing for them.
Remember, all of the ways — fictional or real — that we’ve tried to come up with for testing for consciousness are really tests of processing speed and fidelity. Even the visual tests could be artifacts of our biology.
Except the ability to incorporate a new paradigm — not just regurgitate it, but comprehend it — is beyond the ability of a pattern-recognition system.
Therefore, it is not "I think, therefore I am." It is, instead, "I comprehend, therefore I am."
So back to art. The process of art seems to fulfill this test perfectly; it is synthesis and transformation, which require exactly the kind of comprehension that I’m talking about here.
However, the production of art is not sufficient. In an situation reminiscent of Mead’s philosophy about consciousness and mind, the interaction between art and viewer is separate from the process of its creation. (This definitely applies to written works; I know from personal experience talking to people who have read mine.)
In some ways, the artistic process is a perfect example of both the utility and philosophical frivolity of this whole thought experiment. You can demonstrate that you exist and are conscious through the creation (or transformative appreciation) of art, but you cannot extend that to any other entity.
And as a practical matter, testing for the kind of synthesis and comprehension I’m talking about would be difficult (except when large groups of people self-own themselves). And we haven’t even touched on some of the implications about humans; I’ll just point out that when it comes to children, babies fail the mirror test until about age two, and in some ways they’re operating on the level of rats until five or six years of age. {2}
As a practical matter? I don’t know that it’s going to ever be easily determinable whether or not an entity is conscious — or rather, that they are capable of a self-aware consciousness, let alone if that is their default state.
But, having extended Descartes’ argument sufficiently to handle my own existential question about whether or not I exist, I’m no longer entirely sure it’s the right question to ask.
PS: Yes, the title is a programming/philosophy joke. It made me laugh.
{1} This becomes an even more fraught question when you start talking about multiples and systems, as some people with DID experience it.
{2} From this Radiolab episode, relevant portion of the transcript clipped here.
Featured Image by be*********@gm***.com from Pixabay