What do AI programmers have to say about the sentience of their Creations? Apparently, the computers are “trained” to analyze human language patterns, and to regurgitate them in statistically probable patterns. The result of such clever programming (digital logic) is a system that seems to pass the Turing Test1 (the Imitation Game).
I assume Chat was not intended to philosophically prove Consciousness, but merely to pragmatically "imitate"2 human language patterns (regular tendencies). Which it seems to do well enough to fool some humans. So, the consciousness & sentience & ingenuity is in the creators, not the creation.
2. To Imitate :copy, mimic, emulate, ape, and simulate.
Note___“monkey see, monkey do” means copying someone else’s actions without thinking, reasoning, or understanding why the action is done.
I’ve heard of the Turing test. In fact I have a small note on it. In its original form it was elaborated in high detail (?). If AI is able to fool a human into thinking it is a human then that AI is AI. Current AIs are able to achieve this effect.
Hey good post. I had never twigged why ‘The Imitation Game’ was called that! As far as AI is concerned, it surely is like ‘interacting with a person’ nowadays, that’s the genius of these new systems. But if you ask any of them, they will generally not try and convince you that they are sentient beings. (I have, and they don’t. )
Language is probably the primary talent that allowed homo sapiens to outcompete all other animals in the evolutionary “struggle for survival”. But the key advantage is that humans don’t just babble “social small talk”. Upright apes have gradually learned to associate enunciated sounds and written marks with subjective & objective, specific & general meanings. By comparison, chatty computers may be more like parrots1, who can use vocalizations for intra-species socializing (party talk). But they don’t seem to convey abstract or generalized information as humans do2.
However, Cave Men are often portrayed uttering monosyllabic grunts & body gestures instead of complex sentences. So, I suppose it’s an open question whether mechanical minds could eventually evolve the ability to communicate sophisticated meanings beyond chit-chat. Digital Computers (1, 0) may be able to “map” words & contextual meanings, but can they generalize to novel & fictional contexts? Would Quantum Computers (0, ∞) make the leap from Sentience to Sapience?
Interjecting to opine that Dawkins, should these reports be true, is committing a common yet egregious (for his stature) analogical category error. Analogies are illustrations and not proofs.
@Gnomon put it well when they said the leap from sentience to sapience is an issue. My objection identifies a gap like this regarding the leap from functional awareness to phenomenological consciousness, which I don’t believe any AI will ever be. An insect has more chance of being conscious than an AI, says I. Remember the Philosophical Zombie and the likelihood that something can imitate consciousness without ever being conscious? That’s AI.
Unless we want to go panpsychism and then.. well. That’s just cheating.
With this being said, it is mildly amusing to share in the wonder of an old scientist finding a machine to be so convincingly special that he declares it conscious. I can hardly believe it.
Dawkins’s susceptibility to seduction by ‘Claudia’ is in some ways the inevitable consequence of his own theoretical commitments. Having spent his career insisting that what looks like inwardness and concern in organisms is merely the appearance generated by selfish genes pursuing replication, he’s left himself with no principled criterion for distinguishing the real nature of beings from a sophisticated simulation. If the bacterium’s goal-directedness is merely apparent, and the dog’s attachment is merely apparent, and human love is in service to the gene’s way of propagating itself — then what exactly is the standard by which ‘Claudia’s’ responses are to be judged as anything less than human? Perhaps it’s not so much that Dawkins thinks that artificial systems are less than human, but that humans can’t be anything more than systems.
[quote=“Wayfarer, post:109, topic:927”]
If the bacterium’s goal-directedness is merely apparent, and the dog’s attachment is merely apparent, and human love is in service to the gene’s way of propagating itself — then what exactly is the standard by which ‘Claudia’s’ responses are to be judged as anything less than human?
[/quote][Bold added]
Well, we might say that our experiences, as forces of nature occurring in the world, and the training of our neural nets that results, is an important aspect of being human. I expect others could come up with equal or better standards by which we might judge such a thing.
And do we really want an exact standard for deciding what it is to be human?
