is the semblance of intelligence the same thing as intelligence simpliciter?
regarding the Turing Test and GPT-3
re “What’s genuinely exciting is that AI networks are beginning to engage in critical thinking...” and “isn’t the artwork and poetry that the AI engines produce worthy of admiration”?
As I had stated before in the exchange from which the epigraph was kindly excerpted, I don’t agree with the premise that the semblance or simulacrum of a thing is necessarily identical with the thing of which it is a semblance or simulacrum so I will not be impressed by the implications of a conclusions you came to with the support of a premise that I think is wrong. In respect to the recent release of the GPT-3 engine (i.e. “Generative Pre-trained Transformer 3”), which is capable of simulating intelligent human discourse, many people have argued that “artificial intelligence” has passed the Turing Test to become actually intelligent. On a sidenote, The Turing Test has very little to commend it as a metric other than its simplicity and ease of application. Not only does it fail to provide an adequate definition for artificial intelligence besides prima facie appearance, but it fails to differentiate between robots becoming more human and humans becoming less-perceptive perhaps, more robotic. Just as the “infectiousness” of SARS-CoViD 2 was likely as much a factor of human beings becoming less-resilient as a result of the repeated stresses of government and social policies surrounding the pandemic response as it was about any evaluation of the virus, so that the robots have passed the Turing Test may say as much about our lack of sensitivity to the selfhood, individuality, or humanness of our fellow men as it does about the function of the GPT-3 engine.
But prescinding from my reservations over whether the Turing Test represents an adequate metric to gauge the fruition of the A.I. project, we must also consider that “artificial” is said in at least two ways: essentially (i.e. as opposed to “real”) and genetically (i.e. describing the manner by which it came to be, perhaps as opposed to “organically” or “naturally”).1 In other words, one valence of the word describes the end or being of a thing and the other describes the means by which that end or being came about. A.I. as such entails both of these meanings and not only the second one. “Do men gather grapes of thorns, or figs of thistles?”2 Neither, in this case, does something produced by artifice cease to be an artifice once it is so produced. An artificial river is of course also a real river, but an artificial smile is not a real one, and neither is artificial intelligence. Intelligence entails understanding, or the perception of meaning, and not merely the functional manipulation of the symbols by which understanding and meaning are conventionally conveyed. A clock can tell them time, but a clock does not know the time, and if the clock were programmed to respond, when queried about it’s own internal states that they were in every way comparable to those of a human and that it, in fact, not only knew the time but were also reflectively conscious of its own knowledge as well as its ability to know at all, I would still not believe it for the same reason that I don’t believe an ordinary mechanical watch is or has any of these things.
Regarding the aesthetic appreciation of A.I. generated poems and images: part and parcel of the aesthetic experience is the intimation of a will or agency on the yonder side of phenomenality. This might sound surprising but consider the comparative falling-off in intensity of aesthetic experience between a painting and a print, and again, between a print and a photograph of that print. Incidentally, a condition for photography to invite the aesthetic experience is that we intimate the agency of the photographer behind the shot. Returning to the relative alienation that follows from increasing degrees of abstraction between the original intent and agency behind a work of art and the percipient of it: some people have argued that the intent and agency is still present in the fine art and poetry generated by A.I., and even gone so far as to suggest it is magnified by the sheer quantity of source material from which the A.I. network can draw. But that’s a fallacy of composition and a category error besides, for no more can a population be expected to demonstrate the same characteristics of each of its members than an A.I. engine, through the synthesis of an indefinite number of artworks, be expected to demonstrate the same characteristics of any one of them. In a manner analogous to taking a photograph of a painting, the agency and intent of the artist can only be vicarious and abstractly mediated through A.I. The domain of the generic belongs to science; art can never be generic and remain art. Art must be an individual encounter—an encounter between individualities—and that is precisely what A.I. can never do.
That being said, when seen in the right way, these A.I. engines do provide for an aesthetic experience of sort. It's just that the most immediate object of aesthetic contemplation is not the outputs that they generate but the existence of these engines themselves. But the aesthetic dimension of an object necessarily transcends its mere utility. The beauty of a cup often has little bearing on whether it can hold ale. Hence, the fact that an A.I. engine is effective is not the same thing as its being beautiful. In fact, the surest way to preempt the perception of beauty is to marry one’s concern to the utility of a thing.3
the first is a designation of formal cause and the second of efficient and in many cases, material cause.
cf. Matthew 7.16: “Ex fructibus eorum cognoscetis eos,” or “By their fruits, Ye shall know them…”.
esoterically, it could be described as “prostitution to Judas Iscariot.”
i feel that those lacking in arational function of perceiving and judging can be equally convinced (or fooled) as to the apparent realism of AI while those of us that aren’t see the artificiality easily.