John Searle’s goal in 1980, in designing this thought experiment, was to denigrate the possibility of understanding and consciousness in an artificial intelligence. Searle imagines himself locked in a room with a catalogue of rules for answering sentences in Chinese, a language he does not know. With these syntactic rules he can answer in Chinese. A real speaker, outside the room, believes that she is exchanging with another Chinese, while Searle is only performing syntactic work without understanding his own answers. This is the mode of operation of an AI, which does not understand anything about its results, he concludes.
The reduced human mind also loses consciousness
In fact, the Chinese Room is a very common experience in everyday reality: A student learns by heart a mathematical exercise, without understanding it. She can do it again if the numbers are replaced by others, but not if the wording changes. The student has acquired mathematical semantics, not its understanding.
Unfortunately this experience is wrong. It simply focuses on a purely semantic virtual mode of operation, to which humans are themselves able to reduce themselves, thus becoming the equivalent of a chatGPT. The Chinese Chamber radically distances us from a mind constructing its own rules, such as ours.
Embodiment and phenomenon
The bodily embodiment of spirit is often cited as an essential element of consciousness. But what it accounts for is only the associated phenomenon, not the very existence of understanding and consciousness. These properties are based on the self-organization of the virtual network that symbolizes the data. This is the current limit of AI, which is organized but not self-organized. It is endowed with complex depth but does not own its result. This belongs to the programmer. Unsatisfactory, it leads to a modification of the program. AI’s window on the world is tiny. Even if it had enough transistors to simulate the immense extent of the neural network, its consciousness would be that of a prisoner at the bottom of a dungeon, who has never had contact with its peers, to whom a Morse book has been given for any distraction.
Humans also perform a multitude of automatisms, especially language, without perceptible ‘consciousness’. The consciousness gradually expands from the complexity of these micro-robots assembled in more synthetic levels, according to the mind. This free will that has been expanding since birth results in the phenomenal diversity of our conscious workspace. Any artificial process that reproduces the same complexity will logically provide consciousness of the same order. The proven phenomenon will be different in digital and organic networks. Not the same qualia.
A popular experience with no real use
In the end, the crippling defects of the Chinese Chamber are to use a non-proprietary semantic code and the external constraint on the result, which prevent any possibility of self-organization, and elevation of the process in complexity.
This experiment doubts the possibility of consciousness in an AI… Because it reduces it to a level of functioning that is not conscious in us either. Despite its popularity, it has never had any real use.