Searle’s Chinese Room is horribly flawed.
I would omit it in future discussions of this sort, as it does nothing but muddy the waters and create confusion.
Simply put, either Searle would “understand Chinese” or he would not exist at all (vanishing into an infinite regress of his “Answers to his critics,” where he fails in the unified response, failing to understand that he himself isn’t a robot, and is a system).
Searle’s Chinese Room might be unnecessarily complicated, but I think that it is not nearly as flawed as the “systems reply”.
It is a to begin by assuming that human beings are *nothing but* mechanical systems from the start when that is the very problem that the Chinese Room attempts to dramatize. If the Chinese Room is correct, then rooms do *not* understand Chinese (and neither do ‘systems’ of books, pencils, paper, symbols written on paper, etc) and so there is some additional fact about a mind/person above and beyond the mechanisms that we can observe.
With the systems reply, we can say that Bugs Bunny must be intelligent, since he is part of a system of writers and animators who produce human-like behaviors through a 2D cartoon. Whatever argument we can use to deny Bugs Bunny sentience can also be used to deny a machine sentience.
Think of how a speech synthesizer produces the “B” sound. It has an algorithm based on an abstraction of voltage inputs from a microphone which can be used to control an amplified speaker in the same way. When a human being says “Baby” or “Booboo” the experience may have a similar final output but it is made of feelings and sensations, of lips and exhalation and engaging a voicebox. Our B sounds are further associated with infantile vocalizations, and layers of subtle meaning having to do with nurturing, vulnerability, endearment, etc. The speech synthesizer doesn’t have that and it doesn’t need that to say baby.
Think also the implications of “failing to understand that he himself is a system”. I think this is a fallacy also as it asserts that someone fails to understand their own nature as merely a system, but does not make a case for how such a failure is possible. It is not possible to make a mistake unless there is another option, and if we were only systems, then there would be no other option, and therefore nothing else that we could mistake ourselves for. Machines that make up false ideas of themselves are not necessarily a possibility.
The question that I pose is whether 1+1=2 because it makes sense, or whether our minds make sense because truths like 1+1=2 exist independently of all experience.
If it is the latter, then 1+1=2 stands in for a fundamental set of rules and relations for which consciousness serves to glorify, either accidentally or inevitably.
If it is the former, then that which ‘makes sense’ stands in for a perceptual acquaintance with qualities of undeniable coherence.
It is significant to notice that when we get down to elementary statements such as 1+1=2, we have slipped beneath the realm of logic and numbers without even realizing it. To say that one can be ‘added’ to one and that they are now equal to a group of two is entirely a matter of naming perceptions. There is no real arithmetic going on, we are saying only that when something is to be considered individually we call its individuality “one”, and when we want to consider the presence of one as being adjacent to one other, we call that adjacency “two”. The underlying properties which are being named are conceptually abstracted perceptions. There is no actual “information” named one or two, rather there is a language through which we generalize stereotypical features of our perception - particularly visual and tangible perception. Trying to apply mathematical models to perceptions like flavors and odors is less ‘informative’. They don’t really add up to be enumerable flavors as much as they involve us in a sensory experience in which flavors are both merged and independent.
Lemon + Lime does not necessarily equal two flavors, but can be just as easily thought of as Lemon-lime. Either lemon or lime could be broken down each into multiple flavors including sweet, sour, and citrus, but there remains an idiosyncratic note as well which identifies lemon as one flavor and lime as a different single flavor. Even if we isolate the compounds associated with these flavors, or synthesize artificial compounds with entirely different molecular profiles, there is a huge variation in our perception of any ‘one’ flavor. Lime jelly bean flavor is not the same as key lime pie flavor, yet in another sense, the similarity is self-evident, especially once we give it the name of ‘lime’. It is not a name that is arrived at through a computation or reasoning. Like ‘one’ and ‘equals’, lime is a subjective experience which we can point to but cannot define through a mathematical function.
Does it make more sense, given that the axioms of mathematics as well as physics are defined by subjective expectations (about objective conditions), that we should rule out the idea that all axioms are intrinsically perceptual? We might also ask, if mathematics and information were truly axiomatic, would it be possible to make errors? If our entire conscious experience were made of trillions of precise mathematical reflexes, why is the subject of mathematics even necessary to teach? Wouldn’t it make more sense that we would be able to perform comparatively simple algebras more easily than we can identify whether the flavor of a lime is natural or artificial?